Archives Massachusettsinstitute of Techmfllogy

Home , Rectisol

A COMPUTER MODEL OF THE RECTISOL PROCESS USING THE ASPEN SIMULATOR

Rosalyn A. preston

SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF

MASTER OF SCIENCE IN CHEMICAL ENGINEERING

at the

MASSACHUSETTS INSTITUTE OF TECHNOLOGY

December, 1981

Q Massachusetts Institute of Technology 1981

Signature of Author Signatureof Author Department of Chemical Engineering December 7, 1981 Certified by Lawrence B. Evans Thesis Supervisor Accepted by Glenn C Williams Chairman, Departmental Graduate Committee Archives MASSACHUSETTSINSTiTUTE OF TECHMflLOGY

JUN 1 13982

i IRRA!ES A COMPUTER MODEL OF THE RECTISOL PROCESS USING THE ASPEN SIMULATOR

Rosalyn A. Preston

Submitted to the Department of Chemical Engineering

on December 7, 1981 in partial fulfillment of the

requirements for the Degree of Master of Science

in Chemical Engineering

ABSTRACT

A computer model of the Rectisol process for acid gas absorption has been developed with the ASPEN process simulator. The model uses a modified version of the Redlich-Kwong-Soave equation-of-state to represent the phase equilibria for the methanol- water-aromatics-acid gas system. This report de- scribes the development of the physical properties model and the flowsheet simulation. The work is based on the CONOCO design for a commercial scale coal gasification facility. The computer modelling effort was carried out under contract to the Morgan- town Energy Technical Center, U. S. Department of Energy. TABLE OF CONTENTS

ACKNOWLEDGEMENTS ...... vii

I. INTRODUCTION ...... II. AC:ID GAS ABSORPTION PROCESSES ...... 6 General Features of Acid Gas Absorption . a . 6 Processes ...... · · · 6

Absorption ...... a1 a a 7

Regeneration ...... a a a 7 Solvent Selection ...... * a a 9

Heat Integration ...... * a a 10

Solvent Recovery ...... a . .. 12

Advantages of the Rectisol Process ... . a a 12

Rectisol Process Description ...... * a a 13

III. DEVELOPMENT OF PHYSICAL PROPERTIES MODEL .. . a a 19

Selection of the Thermodynamic Model .. a .. 19

Modified Redlich-Kwong-SoaveEquation-of-State 21

Experimental Phase Equilibrium Data .... . 23

Development of Modified Redlich-Kwong-Soave Model Parameters ...... 25

Evaluation of the Preliminary Thermodynamic Model ...... 27

Enthalpy Calculations ...... 32

IV. SIMULATION OF THE RECTISOL PROCESS ...... 36

Basis for the Process Simulation ...... 36

Absorption Section ...... 38

Flash Regeneration Section ...... 43

Naphtha Extractor ...... 43 Azeotrope Column ...... a a .a 44

Methanol-Water Column ...... 49 iii Hot Regeneration Column .. * O . . .. · 50

Water Wash Column ... . a 0 0 · a a a 52

Heat Exchangers ...... 52

Methanol Make-up . .... 53 a * a a a a a e· Flowsheet Convergence . ··· · 53 V. RECTISOL PROCESS SIMULATION RESULTS.... 55

VI.CONCLUSION...... U a a a a a a 59 APPENDIX A OVERALL RECTISOL MODEL

APPENDIX B ABSORPTION SECTION SIM[LATION

APPENDIX C AZEOTROPE COLUMN SIMUIL MTION

APPENDIX D METHANOL-WATER COLUMN SIMULATION

APPENDIX E HOT REGENERATION COLUMI[ SIMULATION

APPENDIX F WATER WASH COLUMN SIMU LATION

LITERATURE CITED

BIBLI OGRAPHY

iv LIST OF FIGURES

1. The Pipeline Gas Demonstration Plant ...... 2

2. The Rectisol Process ...... 4

3. Comparison of Equilibrium Loading of Chemical and Physical Solvents ...... 11

4. Block Flow Diagram of the Rectisol Process . . . . 14

5. Process Flow Diagram of the Rectisol Process . . . 15

6. Process Flow Diagram of the Rectisol Process . . . 16

7. Henry's Law Constants for Nitrogen in a Methanol and Carbon Dioxide Mixture ...... 30

8. Block Flow Diagram of the Rectisol Computer Model 39

9. Effect of Solvent Rate and Number of Trays on Carbon Dioxide Absorption ...... 42 10. Ternary Diagram for Methanol-Benzene-Water . . . . 45 11. McCabe-Thiele Plot for Azeotrope Column . . . . . 48

v LIST OF TABLES

1. Comparison of Modified RKS Model Predictions with Experimental Data for Methanol-Hydrogen Sulfide- Carbon Dioxide- Nitrogen ...... 33 2. Binary Interaction Parameters for the Modified Redlich-Kwong-Soave Equation-of-State 34 ...... 3. Composition of Feed Gas to Rectisol Process 37

4. Desulfurized Gas Composition . .. 56

5. Acid Gas Composition . . . . o...... 56 6. Regenerated Methanol Composition...... 56 7. Cooling and Heating Duties for the Rectisol Process 58

vi ACKNOWLEDGMENTS

I would like to express my appreciation to all the ASPEN staff who gave me help and support on this project, especially to Herb Britt and Chau-Chyun Chen for assisting me in almost every aspect, to Joe Boston and Paul Mathias for their help with modelling the physical properties, and to Willie Chan and Fred Ziegler for their debugging efforts. I would also like to thank my advisor, Professor L. B. Evans, forhis gui- dance and optimism throughout.

vii I. INTRODUCTION

The objective of this work was to develop a computer model of the Rectisol process (Hochgesand, 1970; Scholz, 1969) using the ASPEN process simulator. This simulation was prepared as part of a program to transfer ASPEN technology to the

Morgantown Energy Technology Center (METC) of the United States

Department of Energy. The overall program involved development of computer models for the major process units in the

CONOCO Slagging Lurgi Coal Gasification process.

The computer modelling effort took CONOCO's design for a commercial scale coal gasification facility as the basis

(CONOCO, 1980). This facility, shown schematically in Figure

1, is designed to generate pipeline quality substitute natural gas from coal. Briefly, the process gasifies coal with steam and oxygen, generated by cryogenic air separation. The gasification takes place at high pressure in a moving bed slagging gasifier using Lurgi technology. Crude synthesis gas is recovered from the top of the gasifier, while the ash is with- drawn in a molten slag form from the bottom. The synthesis gas is subsequently directed to the Rectisol unit for purification. Here, light oils and sulfur compounds are removed from the gas and recovered. Following purification, the gases are allowed to undergo water-gas shift and methanation reactions simultaneously. Carbon dioxide produced in the shift reaction must then be washed out with hot carbonate solution to improve the BTU quality of the substitute natural gas. The final gas treatment involves compression and drying.

1 w 2

ro ad4 ordH a) c-) d E a) 3d for 0 E

o0 EC E Cd o 0 a,

- P .Ha) C t=

- W o 0

p xEC Cd 0 0\ TO

0co 0:z;

e, 0oa)

CD a, a) a) fAt o o 3

The Rectisol process itself features refrigerated methanol as the solvent into which the acid gases in the synthesis gas are absorbed. The process is comprised of an absorption section followed by a series of regeneration processes in which the rich solvent is purified and the absorbed compounds recovered. The diagram in Figure 2 illustrates the particular

Rectisol design proposed for the CONOCO coal gasification facility.

The Rectisol computer model described in this report provides a stream-by-stream heat and material balance for the process. In addition, the model predicts the composition of the outlet streams for a given set of operating conditions.

Therefore, the model can be used to assess the effect of various process variables on the composition of the desulfurized gas and acid gas products. The model also predicts the purity of the regenerated methanol solvent for specific regeneration operating conditions. Separate models of the absorbers and distillation columns yield a rigorous tray-to-tray analysis of each column. Overall, the Rectisol simulation represents a flexible tool which may be applied in future design and sensitivity studies and in trouble-shooting after process start- up.

The input to the simulation is based on the process flowsheet and feed streams from the CONOCO commercial design.

Since Rectisol is a proprietary process licensed by Lurgi Min- eraloltechnik GmbH, the CONOCO design did not provide any information on the actual process operating conditions. For this reason, development of the computer model entailed a 4

o .H 34

i a) Z ) 0 ~o D ts o oo M *H

O 0 Fz ¢ C:

U2 U) W o0 0 Ho a) U2a4 o &4 PE-4 .H 0 V: m X E

E-1

Z 0 5

fair amount of design work to establish reasonable process operating conditions.

The following chapters describe the development of the

Rectisol model. The work initially involved a literature search to locate published process data regarding Rectisol.

Subsequently, considerable effort was devoted to developing a good thermodynamic model of the system to provide adequate physical properties predictions. The actual computer modelling work required testing of several ASPEN models which were still in the development stage. Further, it was necessary to perform sensitivity studies on the design variables in order to establish suitable operating conditions. It was orig- inally intended to use the final Rectisol model to examine the sensitivity of the process to various design variables.

However, lack of computer funds curtailed the project once the model was developed. 6

II. ACID GAS ABSORPTION PROCESSES

This section discusses the general features of acid gas absorption processes in order to provide some background for the detailed description of the Rectisol process. In addition, the discussion serves to present the basic design variables which must be considered in designing a process to remove acid gases.

General Features of Acid Gas Absorption Processes

Removal of acid gases, primarily carbon dioxide and hydrogen sulfide, is important for many industrial plants producing a gaseous intermediate or final product. In the production of substitute natural gas by coal gasification, hydrogen is generated together with a substantial quantity of carbon dioxide.

Since carbon dioxide has no heating value, it must be separated

out of the final pipeline gas to raise the overall BTU value.

The separation may occur at several points in the process.

Removal prior to shift conversion of the carbon monoxide and water to hydrogen and carbon dioxide favors the equilibrium hydrogen concentration. However, since carbon dioxide is pro-

duced in this reaction, further treating to remove carbon dioxide may be required. All sulfur bearing compounds, on the other hand, must be removed before methanation to avoid poison-

ing the methanation catalyst.

Rectisol is just one example of the many acid gas absorption processes. In general, these processes feature an ab-

sorption section in which the acid gases are absorbed into a liquid solvent, and a regeneration section where the acid gases 7 are recovered to yield a lean solvent of sufficient purity for recycle. In addition, the processes usually include a final wash to recover solvent carried over with the acid gas stream. The details of each section are discussed below, highlighting the major factors affecting each section. Absorption

Acid gases are removed from the sour gas by absorption

into a liquid solvent. The gas stream is contacted with the

solvent in an absorber column where it flows counter-current

to the downflowing liquid. The column itself may contain packing or trays, as dictated by the economics of the specific

design. Plate columns generally operate at lower efficiency.

Thus, a packed column design, which requires less height, may

be preferable when a large concentration of acid gas is to be absorbed (Kohl and Riesenfeld, 1979).

Most absorbers operate at low temperature where equilibrium favors absorption. However, the fact that the absorption rate increases with an increase in temperature justifies operating some absorption processes at elevated temperature. Regeneration

The solvent regeneration section serves two functions: namely, to recover the absorbed acid gases from the solvent, and to reduce the gas concentration in the solvent to a low level prior to its recycle to the absorber. A recycled solvent of higher purity will absorb a greater quantity of gas per pass, thereby reducing the solvent recirculation rate.

The tradeoff must therefore be considered between regeneration costs andthe costs associated with a higher recycle rate, such 8 as larger equipment and higher pumping requirements. Further, the purity of the treated gas is limited by the degree of regeneration. The gas will approach equilibrium with the fresh- ly regenerated solvent which it contacts in the final stage of the absorber (Hochgesand, 1970).

There are three primary methods of regeneration. The choice is determined by the nature of the solvent and the desired purity. Flash desorption represents the cheapest method where the pressure of the solvent is let down in several stages.

The final stage is usually atmospheric. However, a vacuum flash may be employed to remove the dissolved gases further.

If flash desorption is not adequate, the gases may be stripped from the solvent following a single flash. Heat must be supplied to the stripper to provide the heat of desorption and to vaporize the stripping medium (often steam). This heat is frequently supplied in the form of steam through the stripper reboiler. As such, the regeneration step can be a significant energy consumer.

Hot regeneration provides the most complete regeneration and consequently leads to the highest purity for the feed gas.

Gas solubility tends to decrease with increased temperature.

This effect is augmented by heating the solvent to its boiling point so that solvent vapors provide some stripping action.

In this case, heat is required to raise the system to the solvent boiling point, to supply the heat of desorption and further to supply the heat of vaporization for the solvent vapors. Although some of this heat may be obtained through heat exchange of the hot overhead gases and hot regenerated solvent 9 with the incoming solvent, heat exchanger losses make external heating necessary (Hochgesand, 1970).

Solvent Selection

A variety of solvents are employed to absorb acid gases.

Despite the varied chemical nature of these solvents, all

should exhibit certain characteristics to render them suitable as solvents. Foremost, the solvent must have a high sol- vency for the acid gas constituents at reasonable operating conditions. In addition, the absorption process must be rever-

sible so that the solvent will release the gas components with

relative ease to yield both gas and solvent of high purity.

It is highly undesirable for the solvent to form an irreversi-

ble compound with any of the feed gas constituents since this may lead to appreciable solvent losses. The solvent should

also have low volatility to prevent excessive solvent losses

in the gas stream by vaporization. Other desirable character-

istics include low viscosity for minimal pumping costs, good

heat transfer properties and low heat capacity to minimize

heating requirements, stability, and low corrosivity. Finally,

low cost and availability are obviously desirable (Nonhebel,

1972).

The acid gas components may be absorbed into the solvent

by physically dissolving or by reacting chemically with an

active solvent component. The absorption process involves

transfer of the acid gases from the gas phase, through the

gas-liquid interface, into the liquid phase. In the case of

a chemical solvent, the absorbed gases then react reversibly to form compounds. Chemical and physical solvents have 10 several distinguishing characteristics which account for the different absorption mechanisms. As the diagram in Figure 3 shows, the equilibrium loading of a physical solvent increases in direct proportion to partial pressure, provided Henry's

Law applies. In contrast, the loading of a chemical solvent rises rapidly at low gas partial pressure, but then reaches a point of saturation beyond which little absorption occurs.

This distinction makes chemical solvents the best choice for purifying gas streams at low gas partial pressures, and physical solvents the choice for purifying high pressure concen- trated streams. The same behaviour dictates the method of regeneration. Physical solvents are easily regenerated by flash desorption, although stripping or hot regeneration may be used for greater purity. However, a saturated chemical solvent subjected to a pressure reduction will release only a fraction of the total absorbed gas. Some form of reboiling is required to effect adequate regeneration (Hochgesand, 1970).

In a chemical absorption system, chemical reactions increase the liquid film coefficient, thereby raising the overall absorption rate (Kohl, Riesenfeld, 1970). Since chemical absorption is more rapid than physical absorption, chemical systems require fewer trays and thus smaller towers.

Heat Integration

Solvent regeneration usually takes place at higher temperature than absorption. As a result, most acid gas absorption processes are highly heat integrated. Often the rich solvent may be preheated and the lean solvent cooled by heat exchange. The absorption process is exothermic so that the solvent 11

Figure 3 COMPARISON OF EQUILIBRIUM LOADING np' rT4.-.MTTAT. ATTh T.TVTAT. (NT.'T/1XTr1Z'

Chemical Solvent -ical e nt Pads I I I I Gas I Partial I I Pressure I I I I I I I I I I I I Pdes I I

:ium

Re ference: Hochgesand, 1970 12

temperature is raised within the absorber. This means that

additional cooling beyond heat exchange is necessary.

Solvent Recovery

To minimize the solvent losses from the system, the gas

streams may be washed with water or other suitable wash sol- utions to recover the vaporized solvent leaving the absorption or regeneration sections.

Advantages of the Rectisol Process

The Rectisol process employs refrigerated methanol as a physical solvent to absorb both hydrogen sulfide and carbon dioxide. According to the literature, the absorption process is carried out at low temperatures, generally between O°F and

-75°F. At this temperature level, both hydrogen sulfide and carbon dioxide are highly soluble in the methanol, although the solubility of hydrogen sulfide exceeds that of carbon dioxide by a factor of 5-6 (Hochgesand, 1970). Methanol has a high vapor pressure compared to most solvents. Thus, the low temperature is also desirable to reduce the methanol losses

due to vaporization.

Rectisol has several attractions supporting its applic-

ation in the CONOCO design. Firstly, Rectisol is ideal for

high pressure processes since the equilibrium loading of a

physical solvent is directly proportional to pressure. In the

CONOCO design, the synthesis gas comes directly from the gas-

ifier and is therefore at high pressure. Further, since Rect-

isol favors absorption of hydrogen sulfide over carbon dioxide, the hydrogen sulfide may be totally removed with only a frac-

tion of the carbon dioxide. Such selective removal is important 13

for two reasons. Firstly, if the sulfur compounds are recovered with a low concentration of carbon dioxide, the gas

stream may be handled directly in a sulfur recovery unit,

such as a Claus Plant, without overloading the sulfur plant

or without recourse to further separation. Secondly, since

carbon dioxide removal prior to shift conversion and methan-

ation is not necessary, utilities would be needlessly consumed

in removing a high proportion of the carbon dioxide at this

stage (Hochgesand, 1970). However, as mentioned earlier, it may be attractive to remove a fraction of the carbon dioxide

on account of the equilibrium considerations associated with

the subsequent water-gas shift reaction.

Rectisol Process Description

This section discusses the specific design of the Rect-

isol process developed as part of the CONOCO Slagging Lurgi

Coal Gasification commercial design. Figure 4 shows a simpli-

fied block flow diagram of the process, indicating the major process sections. A copy of the Rectisol process flowsheet,

taken from the CONOCO commercial design report,is displayed

in Figures 5 and 6. The details of each section are outlined

briefly below.

In the Rectisol unit, the synthesis gas produced in the gasification section is treated to remove the undesirable

sulfur compounds and to recover any light oils in the gas. Purified gas is then directed to the joint shift/methanation section. The removed sulfur compounds proceed to the sulfur recovery unit. C12 0 Iz 10 cil

"Ia) PL,

(a L tr1U2 C C

0 r

I H IIP-)0q XaaE0.CA

H4P-

A_ b4

.0 * Fr

0 U) 15

T r ! '3 . I Wa g. I, . O1' !I _ 5| II_ iia iI I I-

l I ------r------l------t"

1I 1-LI -- , . I I , . -,,I - - -.-r·~ \I -'-"0XIrI X f__

il i - i - I i I I I I I I i I _ .,I _ - ! I I ~ ~~~~~~~~~~I I i I I I I

i :I I i i - I I I r I_ i i i , I -I 1. -, 1-i i I . I I I I 1~~~~ , -- -T-

t r'v"' |Itr . ..-t *lg; (I 5 i *ai I _____ 1 tT7TTh,Iwi '¾ I,'' 4 I (

L----.I- - = .J..

II &' J-, f A1 I ~~~~~~~~~~~'Ia .- i -i I * . ------2I-- A I, " f 1 t i i I .I I , - - I .. I ! ...... t . _ i - . --- fi.

1T I L ,;III , 4 'I

_\ I

- E- .4 --

_ _ I, .t i- 1- - -.-- 4 -, 1- 0 i if ill JA- ! i.. - . .- 0 - .11 4 16

r II i i -, , I I .. g; I .Ii Z,. h8 I. Z z -, 'I .tl~ . I3 i tI i -ill 3: " Z 1 5 ; , I - r ---. .I.t . , _ . I , i -, 11 , 7 til - 211'1 .. ,_ ., I .1 I . is-l I i 9:,1; !?- , _ ., j1 _ 11 tc.;'A I t id i A -- -, i I .;- 1 1,r !;: ,1 . 41. ,'

. '.

i j

3 -

-i ! 3 11

., .g

I I i ; i .I j tI iI i! I ,tf . e ! I

lol~ . \0 it _:.2_i-_ t -i| rt ! a, I' - ,f, ~t4 ... I ( I I I yI I Hl _'_3a , ' . + < F - t ' ----- '

I ,,*i a I .I. I- , ------,t . ...- ; I f,. .. _ --- j w~~_ I L ,I i-- - I, I1 I __ i

,; .,A I Iz - i i -I . f-ts, , I , .F , , i , - j ------, -' I

'IfX ' . ia..

- .ji' -

I _ _ -_ .__~~~___ ~~~=

I - - i --.- __j ! ~----- _ -,-' ------i - , , i. I!I I 14 --i I L.- --- 1 '4 i I! .;I "Ir I I_ .i

1 a I 17

The feed gas is first cooled to absorption conditions by heat exchange with the purified gas and by supplemental refrigeration. The cooled gas then proceeds to the absorption section where the sulfur compounds, naphtha, hydrogen cyanide and water are physically absorbed with refrigerated methanol. The absorption process occurs at high pressure and low temperature in two stages; in the prewash section, all the naphtha, water and hydrogen cyanide are removed with some of the acid gas, while in the main wash section, the remaining sulfur compounds and a portion of the carbon dioxide are absorbed.

The ensuing process steps are devoted to solvent regeneration in which acid gases are recovered from the methanol to yield a lean solvent of sufficient purity for recirculation.

In the first step, the rich solvent streams from the prewash and main wash absoprtion sections are separately flash regenerated. The pressure of the solvent is reduced causing a fraction of the acid gases to desorb.

The liquid from the prewash flash regenerator is sent to the naphtha extractor to recover the absorbed naphtha. Here a liquid-liquid separation is attained on adding the water rich bottoms from the water wash column. The high purity naphtha layer is decanted and sent to storage. The remaining methanol-water-naphtha liquid is fed to the azeotrope column.

This column produces an overhead composed of naphtha, carbon dioxide, hydrogen sulfide and methanol. A mixture of methanol and water is recovered in the bottoms which is subsequently separated in the methanol-water column. 18

The hot regenerator serves to strip the main wash solvent of the gases that were not desorbed during flash regeneration.

The methanol overhead from the methanol-water column is also fed to this unit where the methanol vapors, in conjunction with those generated in the reboiler, act as the stripping medium. The acid gases are recovered in the overhead and combined with the gases from flash regeneration, the azeotrope column and the naphtha extractor. This total gas stream is finally water washed to recover any vaporized methanol prior to being sent to the sulfur plant.

A portion of the regenerated methanol from the hot regeneration column bottoms is used as reflux for the methanol- water column. The remaining methanol is cooled by heat exchange with the naphtha extractor feed and the hot regeneration column feed. After further refrigeration, the regenerated methanol is recirculated to the absorption section. 19

III. DEVELOPMENT OF PHYSICAL PROPERTIES MODEL

Accurate representation of the phase equilibria was essential to the simulation work in view of the lack of process data. Moreover, the physical properties estimation was especially important since the hydrogen sulfide-carbon dioxide- nitrogen-hydrogen-methanol-water-aromatics system is highly non-ideal. The presence of methanol and water, which are significantly polar, and hydrogen, nitrogen, carbon monoxide and methane, which are supercritical, render this system hard to describe. Therefore, considerable effort was devoted to developing a good thermodynamic model.

In order to simplify the system, all the sulfur compounds are treated as hydrogen sulfide, and the C2, C3, C4 compounds are lumped into the methane component throughout this work.

The naphtha in the feed is represented as benzene since coal- derived liquids are generally aromatic. These simplifications reduce the system to the following nine components: H2, CO, N2,

C02, CH4, H2 S, H2 0, C6H6 , and CH30H.

Selection of the Thermodynamic Model

There are two approaches to modelling the phase behaviour of a mixture. The conventional method invokes a liquid activity coefficient model to represent the liquid and an equation- of-state for the vapor. Thus, the system is described according to: iYiPT xf wherei = vapor fugacity coefficient

PT = total pressure 20

Xi = liquid activity coefficient fol= liquid reference fugacity 1 yixi = vapor, liquid mole fractions The alternative approach uses a single equation-of-state to describe both the vapor and liquid phases. Noncondensable components in the mixture present a problem with regard to establishing the reference fugacity for the activity coefficient approach. Generally, the standard state fugacity is that of the pure liquid at the system temperature and pressure.

It is meaningless, however, to consider a pure liquid state for a supercritical component and, thus, it is necessary to follow the unsymmetric convention. In this case, the activity coefficient is normalized to 1.0 at infinite dilution and the

Henry's Law constant becomes the standard state fugacity. As

O'Connell (1977) points out, there are difficulties associated with referencing to the Henry's Law constant for multicomponent solvents since the constant is dependent on solvent composition. As such, the composition dependence of the standard state fugacity should be built into the thermodynamic model.

On the other hand, equation-of-state models have generally proved to be inadequate for predicting the non-idealities of mixtures containing polar substances (Mathias, 1981). How- ever, ASPEN has recently adapted the Redlich-Kwong-Soave

(RKS) equation-of-state to improve its ability to represent systems containing highly polar compounds. Mathias (1981) has demonstrated that this model is effective in describing the coexisting vapor and liquid phases for systems such as methanol-water. An alternative model applicable to polar 21 systems is the Perturbed Hard Chain equation-of-state (Gmeh- ling et al., 1979). In this case, the model handles polar components by assuming that they dimerize. Thus, the model must evaluate the chemical equilibrium for the dimerization process. Such treatment proves to be a major drawback for this equation-of-state since it augments the computational effort substantially. In addition, as Mathias (1981) notes, dimers are rarely observed experimentally, so that the equation rests on a weak theoretical foundation.

Consideration of the difficulties and inaccuracies associated with the activity coefficient model approach led to the selection of an equation-of-state to correlate the Rectisol system for this work. The particular equation-of-state that was chosen was the modified version of the Redlich-Kwong-

Soave model which is described in the next section.

Modified Redlich-Kwong-Soave Equation-of-State

The basic Redlich-Kwong-Soave equation is given by

RT a V - b V( + b) where 'a' and 'b' are functions of critical temperature and pressure as follows

a = (T)a ac= 0.42748 R2 Tci2 P ci

b = 0.08664 RT ci P . ci To some extent, the 'a' parameter may be regarded as an attractive parameter, accounting for the intermolecular attractive 22 forces. On the other hand, the 'b' parameter is a repulsive term, representing the finite volume occupied by the component. c(T) is used to correlate pure component vapor pressure data and has been modified by ASPEN to the following form

05 = 1 + m(1 - Tr05) - p( - Tr)(0a7 - Tr)

Tr = reduced temperature m = function of acentric factor

p = polar parameter

The correlation for m, taken from the work of Graboski and

Daubert (1978), is 2 m = .48508 - 1.55191 - 0.15613ao where = acentric factor.

The ASPEN work has introduced the polar parameter, p, in the expression for to improve the vapor pressure predictions.

This parameter is empirical and must be determined by fitting experimental pure component vapor pressure data.

For supercritical components, Boston and Mathias (1980) recommend a different expression for to extrapolate into the supercritical region

= exp(c (1 - T)) with c = 1 +0.5m +0.3p

d= c- 1 c

Mixtures are represented by the above equations using the one-fluid theory. This means that the parameters 'a' and

'b' become quadratic mole fraction averages with the following form

a - zzx x.a. a = .: ' f1 1i 3 b i xix1 jj.b.. 23

where

0 5 0 1 aijl13ai = (a.a.) (1 - Ka - i K j 1000T )

° 1 b.13 (bi 2+ b i )(1 - K bij -Kbi 1000T ) 2 1 As these equations show, the binary interaction parameters,

Kai and Kb , are expressed as linear functions of the absol-

ute temperature in Kelvin. These parameters account for the

deviation of the behaviour of a binary pair from that predicted according to the basic mixing rules. Any combination of these

four parameters, K K K0 K may be used to fit binary 'a' a' Kb'Kb' vapor-liquid equilibrium (VLE) data to the modified RKS equa-

tion-of-state.

Experimental Phase Equilibrium Data

The physical properties model requires only binary inter-

action parameters in order to describe a multicomponent system.

In the absence of multicomponent vapor-liquid equilibrium data,

values for these parameters may be established by fitting bi-

nary phase equilibrium data alone. Prausnitz (1980) indicates

that his experience has been good in employing only binary

experimental data to predict multicomponent equilibria, pro-

vided that the system is well removed from the critical for

each component. He cautions that, in cases where the critical

conditions are exceeded, multicomponent predictions may be

unreliable even if the binaries are fit well. Unfortunately,

published multicomponent data are not available for the spe-

cific Rectisol system. This deficiency made it necessary to rely on experimental binary equilbrium data to a great extent.

As the work proceeded towards determining the parameters for 24 the modified RKS model, the results of a recent study at North

Carolina State University became available, presenting quaternary data for the hydrogen sulfide-carbon dioxide-nitrogen- methanol system (Rousseau et al., 1981). These data initially served as a comparison for the multicomponent equilibria predicted by the preliminary physical properties model. Subse- quently, the quaternary data were used to improve the binary parameter estimates, leading ultimately to an improvement in the model predictions.

In order to maintain the data fitting effort at a manage- able level, it was initially decided to ignore the interactions between two gas components. Justification for this simplification lay in the fact that the non-idealities of the liquid phase were expected to dominate over the vapor phase non-idealities, as Prausnitz notes (1980). This decision reduced the data fitting to the 21 possible interactions of the gas components with methanol, benzene and water and the interactions between methanol, benzene and water themselves.

The temperatures in the Rectisol process span a broad range, varying from -75°F in the absorption section to as high as 250°F during solvent regeneration. To minimize the degree of extrapolation, the parameter estimation work called for data covering as wide a temperature range as possible. A literature search revealed experimental phase equilibria for each of the binary pairs. Although this information did not provide complete temperature coverage, the data adequately spanned the temperatures where a given interaction was most relevant. For instance, gas-liquid non-idealities are most 25 important at the lower end of the temperature range where absorption takes place. On the other hand, the high temperature processes mostly involve the separation of benzene, methanol and water.

In cases where more than one data set were located for a binary pair, only the most recent was selected for the model development unless these data appeared to be generally incon- sistent with previous data.

Development of Modified Redlich-Kwong-Soave Model Parameters

Fitting the modified RKS model to binary experimental data involved adjustment of a combination of the four interaction parameters, K, K, K, K1, as discussed previously.

The best estimates for the combination of parameters were generated for each set of binary data using the ASPEN Data

Regression System (DRS). DRS employs the maximum-likelihood principle in reducing experimental data. This analysis recog- nizes that experimental data contain random measurement errors which it takes into consideration, assuming that the errors are randomly distributed about zero with a normal distribution.

The standard deviation associated with a particular measurement is supplied by the user. The system creates an objective function comprised of sum-of-squares terms for each observed variable, for example: N e 2 o 2 o. e 2 (Pi-Pi) -T i ) (Xlil) T. + . + 1 a,.2 x2 P, Cr1 1 .1 in which the superscript 'o' stands for the value adjusted according to the thermodynamic model, while 'e' refers to the 26 experimental value. The terms are summed over all data points.

In DRS, the Britt-Luecke (1973) generalized least-squares al- gorithm computes estimates for the model parameters such that the objective function is minimized. At the same time, the parameters must satisfy the equilibrium constraint that the fugacities of all equilibrium phases are equal.

Prausnitz (1980) presents an informative discussion of the methods of estimating parameters. Many of his recommend- ations were adopted for this work. For instance, the following values were used for the standard deviation of the measured variables, as Prausnitz suggests: temperature - 0.050°C, pressure - 1 mmHg, liquid mole fraction - 0.001, vapor mole fraction - 0.003.

The data regression work demonstrated that, in most cases, two parameters adequately correlate the equilibrium data. How- ever, in the final set of interaction parameters, additional parameters are included for some pairs if these improve the fit significantly. One notable example is the benzene-water system for which all four parameters are essential to modelling the observed liquid immiscibility. It should be noted that, in order for the 'a' and 'b' parameters in the fundamental RKS equation to retain their theoretical significance, the term

(1 - K0 -K1 T ) and the analogous term for the 'b' para- aij aij 100-0 meter must have a positive value. Unfortunately, in the case of benzene and water, large parameter values are necessary to represent the equilibrium, rendering the above terms negative.

While this anomaly causes no computational difficulty, it does mean that the model must be considered to be purely empirical 27 as far as benzene and water are concerned. As such, there may be some concern about using the equation-of state to extrapolate information. However, the benzene-water experimental

data adequately cover the temperature range of importance in the Rectisol process. Further, a comparison with experimental

data demonstrates that the model provides a good prediction of the liquid-liquid equilibrium for the methanol-benzene- water system. This comparison is illustrated in the follow-

ing chapter.

It has been suggested that highly non-ideal mixtures,

such as the benzene-water mixture, may be better represented

using the recently proposed mixing rules based on the local- composition concept (Boston, 1981; Whiting, Prausnitz, 1981).

Further work in this area lay beyond the scope of this project.

In general, good agreement was obtained between the experimental binary data and the thermodynamic model predictions.

The maximum error in the liquid composition for a set of bi-

nary data was, on the whole, less than 6 percent. For a given

binary pair, the binary parameters exhibit a high degree of

correlation which indicates that the parameters are not uniquely

determined. In other words, the experimental data may be rep-

resented by several alternate parameter combinations with

little loss of accuracy.

Evaluation of the Preliminary Thermodynamic Model

The preliminary thermodynamic model of the Rectisol

system was comprised of the modified RKS equation-of-state

with interaction parameters generated by regression of binary 28 data, as just described. This model neglected any gas-gas interactions, considering them to obey the simple mixing rules without correction. To assess the validity of the model, the predicted equilibria for the hydrogen sulfide-carbon dioxide- nitrogen-methanol subsystem were checked against the data presented by the North Carolina State University team (Rousseau et al., 1981). Bubble point calculations using the experimental liquid composition and temperature permitted a comparison of the predicted bubble point pressure and vapor composition with the actual data. Unfortunately, the model did not prove to represent the data adequately, overpredicting the bubble point pressure by as much as 30 percent and overestimating the nitrogen in the vapor at the expense of the carbon dioxide. In an effort to ameliorate the situation, parameters modelling the gas-gas interactions (carbon dioxide-nitrogen, hydrogen sulfide-carbon dioxide, hydrogen sulfide-nitrogen) were regressed from binary data and included in the model. The resulting

improvement in the model predictions was not considered sufficient. Further, the predicted Henry's Law constants for nitrogen in mixed methanol-carbon dioxide and methanol-hydrogen

sulfide solvents exhibited unexpected behaviour. The experimental data appeared to be consistent with the approximate relationship that

ln HN~~, mix =Zixi ln HN, i(Prausnitz, 1980) with x. = solvent mole fraction on solute free basis 21 However, rather than displaying linearity on a logarithmic basis according to the above expression, the predicted Henry's Law constants increased initially with a small increase in the 29 amount of dissolved carbon dioxide or hydrogen sulfide. The constants appeared to reach a maximum before decreasing to the

Henry's Law constant for nitrogen in pure carbon dioxide or hydrogen sulfide. Figure 7 illustrates the variation of the

Henry's Law constant with solvent composition. The plot includes three curves. The upper two demonstrate the improvement in the behaviour of the constant on inclusion of the gas- gas interactions. The lower results from utilizing the quaternary

data in estimating the parameters for the physical properties model, which is discussed in detail in the following section. The abnormal solubility behaviour is in part re- sponsible for the deviations between the predicted and experimental compositions. In the model, a small amount of dissolved gas in the methanol results in an unexpectedly high Henry's constant and consequently a lower solubility for nitrogen.

Therefore, the vapor mole fraction of nitrogen is overpredicted.

From an analysis of the experimental data and the model predictions, it was concluded that the vapor interactions

between carbon dioxide and nitrogen are strong and conceal any non-idealities in the liquid. This fact makes the system hard to analyze. It did not appear possible to isolate the vapor and liquid behaviour from binary data using a single equation-of-state. The apparant success of Rousseau et al.

(1981) in modelling the quaternary system may be attributed

to the fact that they adopted the activity coefficient ap-

proach. -This means that they employed a separate equation for each phase. In their model, the vapor is represented by

the RKS equation-of-state with interaction coefficients set 30

0 ;

C0 0 ~: 0 E-q H Z X0 i--.tH0 H 0 ,

.,io*H P4zO Z 00 Oi00 *H Cl. EOZm z o . .H H I--a) *0 r- ¢CD Qa4 H0 > E1 --I Z~v () N < 0

0 O\ 0 0 0 0 0C) C- 0 0 0 00 0 0 0oo 0 0 V 0 CV 0 1,2_) n cn CM -A- omq'e'ueosjqTN oJ us!suoo M 1j sRJuaH 31 to the standard vapor values available from separate vapor phase studies. Having established the vapor behaviour in this manner, the liquid behaviour could be considered independently.

In working with the modified RKS equation, it appeared that the 'a' parameter, the attractive term in the basic equation, could be treated as the vapor term, while the repulsive 'b' term could be considered as representing the liquid. This observation suggested that the vapor interactions could be isolated from those in the liquid by fixing the interaction parameter for the 'a' term to the standard RKS vapor value reported in the literature. The remaining non-idealities due to the liquid phase could then be reflected in the 'b' parameter. It should be noted that this approach is only suitable .for subcritical components since the modified RKS equation only reduces to the original RKS equation at conditions below the critical.

For the final model. it was decided to use the quaternary experimental data in the data regression runs to generate improved parameter estimates for the interactions between carbon dioxide, hydrogen sulfide and nitrogen. So that the

DRS runs would converge, several parameters were fixed before

the final DRS run rather than fitting all the gas-gas interactions with the quaternary data. Hydrogen sulfide-nitrogen gas interactions appeared less significant, so the original parameters developed from binary data were used. The effect of the vapor interactions for the hydrogen sulfide-carbon dioxide pair, which are both subcritical, was modelled with the 32 standard RKS value for K (Graboski and Daubert, 1978). Data a regression of the quaternary data then established parameter estimatesestimates for aKtrgend Kb for carbon dioxide-nitrogen andand ththe liquid K for hydrogen sulfide-carbon dioxide. The resultant model predictions agree substantially better with the experimental data as the comparison in Table 1 shows. The bubble point pressure predictions still appear to deviate from the data, especially at the higher pressure levels, however.

Table 2 displays the final parameter estimates, indicating the literature source of the data. This table shows that the final set includes values for several additional gas pairs which were readily available from previous work (Mathias, 1981).

The parameter set is stored in a user library under RKAKIJ, to be called directly by the simulation program.

Enthalpy Calculations

In order to guarentee reliable enthalpy predictions from the modified RKS model, enthalpy data should be included in the data regression runs to determine interactions parameters.

However, such data were not available for the Rectisol system.

Under these circumstances, the Lee-Kesler equation-of-state was favored over the modified RKS equation-of-state for enthalpy calculations. The former correlation has a good reputation for predicting reasonable mixture enthalpies based on the pure component properties. 33

Nc O N N N C V O, 0 ~ 0 O\ V I\0 c- c- - \O O' O'- O~ o O co x -0i-0 co 0i 0 0 c'- CO \0o 0 0 0 T- - O 0 0 0 0 o 0 0 C;O C; O C~O O;' U O C' O Co 0 0 0 0 n n N On o 000 o o O O O O oO 0 O 0O 0 O NC o;0 n 0Cn \0 o \) 0 c'- O 0 0 0 q C0 0 0 0 iN i- NM n O 0 C\l n 00 0 0 ~. V- --I n - w n m m) 0 C O n CO N CO cO 0 0 0 o 0 N n o x * * 4 wr6 IW l ;__>T I' * i.1% ad) E-4M ~ cO o # u -4 i2CH o 0 * No0 0c N 0 o N N N O~EV2 ~ O O O O o 0 0 0 ,C- \o0 C)-o 0n :HE0 0 0 0 Io o. o o s a 72 N 0 0~C \0 O¢ C TO CO CO N o fail 0 H tz H MN CoiH 0N \H t0C" e 0 " 0co 0 ii ov- CNo T- o CMlo UE0 S\ 0 0 0 0 * * H r z0 \PO 0O 0O 0O 0 C'2P ¢ X O0 0 0 0 0 0 0 co \0 C \0 E Eq CM0 O O o~ o C C'- V3 NO~~~- 0N 0 O OCO 0 0o 0o 0 0 O O O O o 0 0 0 O o 0o o ~:2~ O O O O o o o o O O 0 O o o 0 0 _-

o~-; o; I4(O c C On -000

a) c .. 0s 0 CY' ON,Iq._ N,-4 NON, CnO O O O O So CQl C\1 Cr 0 0 0 0 I I I I 34

C-cr\ O ON ON \ \n o7\ C\ U--U) C- C.'- ON\_ -CO -C'--'\ * Cd * a) . -a)OO a) c~~- C~ Cd cd - cd~ Cd Cd N N N Nc N N *H ONON a) 3 ON O\ 4-' 4- +- O O, r O\ O\ O\ O\ 4- a) . i-I # d i-i i-I ~-I i-I Cd v-f o a) O ON a) H - Cd ' U2 _ -a) a) S Ldd Cd~ d _ O CO ') a) -P C HHH H ) . ) )HHr- 0 ) ~> - O -P a a) cN N U a) a) a) cd a) a a) U) cqcd $s Cd 4S .- *H *H 4-' 0 0 *H H bfl * O rO ;4 rO rO' C = - *H Mq -M -, M 0 c) Cdo w a) 00 i 000-PHM fq (~: >-i >-i 11

H H oN \ O 0 N 0 C- C- V, 0 N H \ ON\ CO E C.- O0 D C O O· -gj -~m~- CY0?C N -.ION 000N\O ONH - NO0 0 I I I I EI cnC C nN .- coCO \ T-0 [,--' CcOO O O \ Eq D 0 ~ 0 0.4 .-. 0 0 O;Z 0", 0 -.I~ \D C\? C>1 01ON coNC ON CO 0 \ CO N V)' c O N 0 v-)0 0 O 0o Ui)- 0 0 O C')\00 °c' co V-4 V) V' H tr> 0000 o000 000000 0000 I V' V)

~HZ C-\o tq:E 00 0 0 -. o0 0 0 0 0 0 o 0 (O O (O O A ( (H (O (O (O (O (OO (: O ( od-t I I \0 cO I I n C\l C\ O

*,r Cd P-4 0 N = U N CM C\lNe -' \0 cN NO= NO X: :N \ NO NO X: CN CN 0 CZ CN C, = VZ 2 C.V C. M 0 X V Z C.) V :M: OV t) Z = a)Z 0 p 0S C- :m = t' m: = m 0000000(n n n n n n n 0 0 0 0 0 0 ,0 = \0 ' \00 , 0 00000000V V : V = V = 0V V = V : =:rm = N =: N mt N0 0V O0 V 0 V\ rO) 35

aNC\ o O U

U~ V~ \O\0QJ cO ON G\\Q '-4 N-i a) U: OOH -P CC -) · \ l Cd o , -4 a) la) H *H +' _ a) H H -P N 0 a. E - o · H r o 0 oi o 0 *H A >0-4 U2o E- S ) 0 H o - C\ 1-I H O\ co oD O q Eq -.I-C\ -- 0r E-1V2 Oocl-0 0o 0 0 m -0C\l' '~~~~~~0 I- 0 I 0 oI o0 o C\.l 1Lf\Cn o0 PI-I'H - CWl n V) U E- C\Z o I I I 0o00· o o a) 1-4 I I I EH Z0 I'Z 0- wE

O V2i ROO cc ooc0 o o 0 EH 0 o o 0 O · o 0 o 00000a a a

CO\co Z H, \0 ON H 0\.O 000 R'- 0 00- 00 0 0\ i cs o I 0 H~ I

*ri

U) _ N -p Oj~4=:a: C~l 0 O O 0C0 O a) 0 0 V

0 0 C\ 0 C'JeC~io 36

IV. SIMULATION OF THE RECTISOL PROCESS

Basis for the Process Simulation

The proprietary nature of the Rectisol process restricts

the amount of process data disclosed in the literature. The

information presented in the CONOCO Commercial Design Report

consists solely of the feed and product compositions and a process flowsheet. While the process flowsheet shows detail

of the sequence of operations and the destination of all in-

termediate streams, no specifics are provided for the process

operating conditions. On account of the lack of actual oper-

ating conditions, it was impossible to develop a simulation

that would match the CONOCO design. Therefore, this work

adopted the approach of developing a viable Rectisol design

based on the CONOCO design flowsheet and feed composition.

The feed gas composition, pressure and temperature displayed

in Table 3 were taken for the process model.

The literature search located typical absorption temp-

erature and pressure ranges only. Detailed simulation of the

other process units using ASPEN established the remaining

operating conditions necessary to achieve the desired separ-

ation.

To minimize the computer time and cost associated with

simulating a complex cyclic process such as Rectisol, a sim-

plified process model was assembled. In this simulation, a

series of simple unit operations such as mixers, splitters,

flashes, heaters and pumps represent the Rectisol process. Rigorous simulation of the staged separation units is performed 37

Table 3

COMPOSITION OF FEED GAS TO RECTISOL PROCESS

Component lbmol/hr Hydrogen 1909.7 Carbon Monoxide 4052.9 Carbon Dioxide 299.1 Methane 471 1 Ethane 18.9 Ethylene 2.2 Propane 4.7 Propylene 2.2 Butane 1.8 Butylene 1.6 Nitrogen 41.7 Hydrogen Sulfide 131.0 Carbonyl Sulfide 8.0 Carbon Disulfide 0.1 Thiophene 0.4 Hydrogen Cyanide trace Naphtha 23.1 Water 13.8

Total 6982.3 Temperature, OF 95 Pressure, psia 427 38 on a stand-alone basis, external to the overall flowsheet.

The results of such rigorous simulations established the splits to be used in the simplified model, where the units are modelled by separator blocks. A block flow diagram of the overall flowsheet simulation is shown in Figure 8, indicating the correspondence between the model and the actual design flowsheet. Appendix A contains the computer input and output for the simplified Rectisol model.

The development and design philosophy employed in each of the major process sections is discussed in detail below. Absorption Section

The hydrogen sulfide absorber consists of two separate sections; a prewash section, which removes all the water and naphtha from the synthesis gas, and a main wash section, where the bulk of the hydrogen sulfide absorption occurs. Each section was simulated independently using RADFRAC, which is ASPEN's rigorous model for multistage vapor-liquid fraction- ation. The results of these simulations were used to set the degree of absorption in the overall flowsheet simulation in which two mixers and two separators (PREWASH1, PREWASH2, H2S-

ABS1, H2SABS2) model the absorption section.

In designing the absorbers, the designer must establish the number of trays, the temperature of the entering gases and solvent, and the operating pressure. For a given degree of absorption, these parameters then dictate the required solvent rate. Hochgesand (1970) indicates that for the Rect- isol process the methanol circulation rate greatly affects the process economics, since it influences the size of all process I i 0 239 -sp

-1-- - -- ......

~1 &| s--3--- -- g I'

i I

------TI

I '( I

-,,1

CL_r1 . ' ? w 8 3

i III

W 2 ! . ! , I = a' -.-I

41 o I 11

I- t I F 4t1 1

A . I~~~~~1

-4 I a ...... ; X/.... X7-- .1 I p I,! Lh .0 I W--- j O CD 4 Ge~~~~~~~~~~~~~~~~~~~~~~~~~~~O QC ~ X ~i - ~~~~~~~I I ¢- P

Absorption columns may contain packing or trays, although there is some evidence in the literature to suggest that trayed columns are more common for Rectisol (Maddox, 1977; Herbert,

1956). In the current design, the absorbers are designed as trayed columns. The pressure drop through the column is based on the rule-of-thumb of 0.15 psi pressure drop per stage (Treybal, 1968).

The design for the prewash absorber features six theoretical trays and a solvent rate set at 25 percent greater than the theoretical rate required to absorb all the naphtha. This follows the recommendation of Kohl and Riesenfeld (1979). Given the objective of the prewash absorber to recover the

naphtha with as little hydrogen sulfide as possible, the meth-

anol rate is restricted to the minimum that ensures complete

naphtha absorption. In order to improve the performance of

the RADFRAC model for the prewash absorber, the feed is pre-

flashed so that any liquid may bypass the absorber. This

preflash simulates what would actually occur in the bottom

of the absorber.

The gas from the prewash absorber proceeds to the main

wash section where all the hydrogen sulfide and part of the

carbon dioxide are removed. The methanol solvent rate to the

main wash section was determined so that the degree of carbon

dioxide removal agreed with the information presented in the

CONOCO Commercial Design Report, while achieving complete removal of hydrogen sulfide. However, the basis for the ab-

sorber feed composition did not account for the small amount

of recycled hydrocarbon gas. This means that the predicted

carbon dioxide gas removal in the overall simulation is not

in complete agreement with the CONOCO commercial design.

It is unusual for an absorber to contain less than six or more than ten theoretical trays (Perry and Chilton, 1973).

A sensitivity study to the number of trays in the main wash

absorber indicated that an increase from six to ten trays re-

sults in a reductions of only 2 percent in the required meth-

anol rate. Figure 9 shows a plot of carbon dioxide in the outlet gas versus solvent rate and number of trays. The rel-

ative insensitivity of the required amount of solvent to the 42

Figure 9

EFFECT OF SOLVENT RATE AND NUMBER OF TRAYS ON CARBON DIOXIDE ABSORPTION

260

d~ 220 02 in Feed 0o )er

d 180

;C\2 140 ro0- 0 O l . D- 3000 3400 3800 4200 4600 5200 Methanol Solvent Rate, lbmol/hr 43 tower size justified designing the main wash absorber with six trays.

The output from simulating both the prewash and main wash absorbers rigorously is included in Appendix B. In the final simulation, the main wash and prewash absorbers operate with

3100 lbmol/hr and 330 lbmol/hr of circulating methanol respectively. Flash Regeneration Section

In the flash regeneration section, a fraction of the absorbed gases are recovered from the rich solvent by a reduction in pressure. The pressure of downstream processes dictate the final flash pressure. In this simulation, the prewash flash regenerator flashes the prewash solvent to 30 psia, allowing sufficient pressure head to feed the water wash column without further compression. Similarly, the main flash regenerator is designed to flash the main wash solvent to 30 psia in two stages, with an intermediate pressure of 75 psia. Since the vapor from the first flash is recycled to the absorption section, the intermediate pressure must minimize the amount of hydrogen sulfide and carbon dioxide desorbed while recovering as much of the desired hydrocarbon gases as possible. An intermediate pressure of 75 psia represents an appropriate compromise. The three flashes are simulated as two-phase adiabatic flashes (PRWSFLSH, FLSHRG1, FLSHRG2). Naphtha Extractor

In the naphtha extractor, the methanol-water-naphtha-gas stream is allowed to separate into two liquid phases to permit removal of the naphtha-rich phase from the system. The feed 44 consists primarily of the bottoms from the prewash flash regenerator to which a water-rich stream is added to effect the liquid-liquid separation. The water source is the bottoms from the water wash column. In order to ensure that the naphtha extractor operates without any of the water icing, the prewash flash bottoms is preheated to 50°F. The design pressure is 30 psia, 5 psi higher than the water wash column to which any vapors are directed. At these conditions, the physical properties model predicts a phase separation into two liquid phases which are in reasonable agreement with liquid-liquid experimental data for the methanol-benzene-water ternary. The plot in Figure 10 illustrates the comparison between the predicted separation at 38°F and experimental data at 77°F. Unfortunately, published data were not located at the actual separation temperature.

In the overall model, the naphtha extractor is represented by a feed mixer (NEXMIX) followed by a three-phase adiabatic flash at 30 psia (NEXSEP). Azeotrope Column

The objective of the azeotrope column lies in removing the residual naphtha and dissolved gases from the methanol- water mixture. Preliminary simulation of the column using RADFRAC suggested that a good separation is possible in this column, yielding an uncontaminated methanol-water stream.

This simulation indicated that all the gas components and naphtha are distilled in the overhead along with a small portion of the methanol, while all the water and remaining methanol are recovered in the bottoms. As the name of the d) a) 45 N a)

wIu 1 9I4 1-1 M a) I C 46 column suggests, the overhead may be expected to contain benzene and methanol in their azeotropic compositions (61.4 mole methanol at 128.8°F). Further simulation of the column with

RADFRAC demonstrated that in the overhead methanol and benzene may indeed reach their azeotropic composition on a methanol plus benzene basis.

Development of a feasible design for the azeotrope column involved setting the distillate rate, the reflux ratio, the number of theoretical trays, and the feed entry point.

Seeing that the split between the overhead and the bottoms was well defined in the preliminary study, the distillate rate could be established purely by material balance considerations. The behaviour of the column proved to be highly sensitive to the reflux ratio. The amount of material in the rectifying section of the column depends directly on the reflux ratio. Since such a large percentage of the feed (94 mole percent) ultimately appears in the bottoms product, a high reflux ratio is necessary to ensure adequate vapor flow throughout the column. In the final design, a reflux ratio of 2.0 leads to satisfactory operation. While this ratio appears to be substantially higher than the minimum reflux ratio, difficulties were encountered in obtaining a converged RADFRAC solution at lower reflux ratios.

A revealing method of interpreting tray-to-tray output uses the idealized McCabe-Thiele techniques (Russell, 1981).

This tool provides valuable insights despite the non-ideality of the particular system. If methanol and benzene are identified as the two components which require separation, and 47 all the compositions are normalized on a methanol plus benzene basis, a McCabe-Thiele x-y diagram may be prepared. Fig- ure 11 displays such a diagram for the upper portion of the azeotrope column, where compositions are plotted on a logarithmic scale for clarity. The equilibrium compositions were taken directly from the RADFRAC output. This type of plot identified pinch points, the minimum reflux ratio, and indicated the optimum feed location during the column design. In the final design, the feed enters at the third stage. Figure

11 suggests that a slight improvement in the column efficiency may be realized with the feed located at the second stage.

The diagram also shows that the physical properties model predicts an azeotrope composition close to that reported. Since the overhead lies at this composition, it can be seen that increasing the number of trays would not improve the separation.

The feed quality plays a significant role in the column design, as well. Given the fixed distillate rate of 45 lbmol/hr a simplified material balance analysis, assuming constant molar overflow, predicts the maximum percentage of vapor permitted in the feed for the column to operate as desired. For instance, with a reflux ratio of 2.0, the feed must be greater than 82 percent liquid in order to meet the desired distillate rate. Higher degree of vaporization results in insufficient vapor in the stripping section. In the actual process flowsheet, the azeotrope column feed is preheated by heat exchange with the waste water from the methanol-water column. The final temperature of the waste water was taken to be 118°F, as 48

i Nj i U-- I- I :: 0 !15, i . , !, i mm+mwI i NJ I 1 1 - I I - i . I I; I I i i ' i I I ' Io

A , . 'CA ¢

, . I I I - 0 N

o0 4t.. b

E0: aio 0 4

0 J. H¢C) N: a) O 0 x4 rG N _. \0 4z I~~~~-,e7 _ J H- -I I is \0 I ; e I I, XO Eq4 .. .$ , j . - .Ii: ! I ; . m 4ffll t.? Ai' I -_ I !---- . I I"t~ I I I It I t CD _ I T . I ¢ f- . E

-[ - - -

IL 4~D I I I1

1-H0 H +Hg

_ _

I. .11 I : 'li. I' 1 I ' - I · i i i i , I I I i i i i i 2 0i I I i i I i i i .! I I I i Ill : I N 10-g D P 0 1D, mq N (2) ) q o , , o In I IvWg c %t 0 - 0 0' 3a10Y ,l 49 indicated on the CONOCO flowsheet, for this simulation. This heat exchange results in an all liquid azeotrope column feed at a temperature of 96°F.

The rigorous simulation of the azeotrope column is shown in Appendix C. This simulation required 40 outside loop iter- ations to achieve a converged solution. The difficulty in converging is a consequence of the non-ideality of the system in which the outside loop variables for RADFRAC, namely K- values and enthalpy, are highly composition dependent. Methanol-WaterColumn

The methanol-water stream from the azeotrope column bottoms is separated in the methanol-water column. In this simulation, the column is designed to yield a 99 percent pure methanol distillate. A RADFRAC model was used to develop the column design, and to examine the sensitivity to various design parameters. In addition, a McCabe-Thiele diagram provided initial estimates of the number of stages required to effect the desired separation at a given reflux ratio. This same diagram showed that the minimum reflux ratio occurs at a value of approximately 0.9.

Perry (1973) indicates that setting the reflux ratio 30 percent higher than the minimum reflux ratio generally mini- mizes the combined operating and fixed costs. This general rule was followed in chosing a reflux ratio of 1.2 for the column design. At such a reflux ratio, the desired methanol- water separation requires 25 theoretical trays. The final design introduces the feed on the nineteenth stage from the top which is the optimal stage according to the McCabe-Thiele technique. 50

In the RADFRPAC model of the column, a total condenser is included on the overhead. In the actual flowsheet, however, the overhead vapors proceed directly to the hot regeneration column to act as reflux. Since the methanol overhead and the hot regeneration column bottoms differ very little in composition, the hot regeneration column is, in essence, acting as the total condenser.

Detailed output from the RADFRAC simulation is located in Appendix D. In the overall flowsheet, the column is represented by a mixer and separator block (MWMIX, MWSEP) with the reflux rate calculated by the Fortran paragraph

REFLUX.

Hot Regeneration Column

The purpose of the hot regeneration column is to strip the main wash solvent completely of the acid gases remaining after fash desorption. Methanol from the methanol-water column is also fed to this column. The methanol vapors, combined with those generated in the reboiler, strip the acid gases from the solvent.

The column is designed with 10 theoretical stages with the main wash solvent and methanol distillate feeds located on the third and sixth stages from the top, respectively.

The column was simulated with RADFRAC at reflux ratios ranging from 0.5 to 2.0. These simulations indicated that it is possible to strip essentially all the acid gases from the methanol at reflux ratios of 1.0 or greater. At the same time, the overhead vapor contains less than 0.07 mole percent methanol vapor. Increasing the reflux ratio from 1.0 to 2.0 does not 51

improve the separation significantly. However, at a reflux

ratio of 0.5, the separation is inadequate, resulting in approx-

imately 13 mole percent methanol in the overhead. The overhead

column temperature varies greatly with the reflux ratio. At

ratios exceeding 1.0, the overhead contains practically no methanol and must be cooled to -40°F or less in order to

condense sufficient material for reflux. The temperature is

low enough so that the acid gases require no further refrig-

eration prior to the water wash column. On the other hand,

at low reflux ratios, the overhead may contain a significant

percentage of methanol vapor and, therefore, the column can

be refluxed by condensing the methanol. For instance, at a

0.5 reflux ratio, methanol accounts for 98 percent of the

condensate which is condensed at 97°F. Under these circum-

stances, a significant amount of the methanol is lost with the acid gases. Some of this may be subsequently recovered in the water wash column. Thus, the design of the hot regeneration column must consider the trade-off between a high condenser duty requiring refrigeration on one hand, and increased methanol losses on the other. Such a trade-off lay beyond the scope of this work. For the purpose of this simulation, a reflux ratio of 1.0 was selected where the methanol losses are negligible.

The detailed RADFRAC simulation of this column established the splits for the separator (HTRGSEP) representing the column in the overall flowsheet. The RADFRAC model is included in Appendix E. 52

Water Wash Column

Methanol vapor is recovered from the combined acid gas

streams in the water wash column. The water-rich stream from

this unit is sent to the naphtha extractor to effect the liq-

uid-liquid separation. The rate of water addition was determined from experimental data so that the subsequent liquid-

liquid separation in the naphtha extractor would yield a methanol-water phase containing less than 1 mole percent benzene.

The results from simulating this column as a four stage ab-

sorber with RADFRAC (see Appendix F) demonstrate that such a

water rate is sufficient to ensure no methanol carryover with

the acid gas. In the simplified flowsheet, the water wash

column is modelled with a separator block (H20WSH) which sep- arates the feed according to the distribution predicted by the rigorous model. Heat Exchangers

The naphtha extractor feed/regenerated methanol, azeotrope

feed/waste water, and hot regenerator feed/regenerated methanol heat exchangers are each modelled using two heaters. The heat duty calculated in one heater is passed to the second heater, which represents the other side of the exchanger, as an input

stream. The split in duty between the two regenerated methanol coolers is determined so that the naphtha extractor feed (prewash flash bottoms) is preheated to 50°F to ensure operation of the naphtha extractor without any icing.

The absorber feed gas cooling train is handled by a series of heaters. The simulation does not model the detailed heat exchange with the purified gas. An enthalpy balance, however, 53 indicates that the heat removed from the feed gas in cooling the gas to -30°F is sufficient to heat the purified gas to

67°F. The CONOCO Design Basis shows that the purified gas is heated to 77°F after heat exchange. Thus, the absorber feed temperature may be achieved through heat exchange with the purified gas without any supplemental refrigeration.

Methanol Make-up

A methanol make-up stream contributes to steady operation of the hydrogen sulfide absorbers by maintaining the methanol circulation at the design rate. The make-up stream replaces any methanol lost from the recycle stream. In the simulation, a Fortran paragraph (MAKEUP) calculates the required make-up rate.

Flowsheet Convergence

To facilitate the convergence of the overall flowsheet which contains several recycle loops, the regenerated methanol recycle stream is broken. More specifically, the methanol solvent to the absorption section (RGMETC) is treated as a feed stream, while the regenerated methanol from the hot regeneration column (RGMETC2) is considered as a product stream.

This simplification is possible since the stream composition

is well defined and the stream rate is maintained at the de-

sign rate by the methanol make-up stream. Such treatment permits the flowsheet to be divided into two subsystems which can be converged separately.

The flowsheet is converged with an ASPEN generated tear stream set using the Wegstein convergence method. Three 54 material streams and two heat streams must be converged to converge the entire flowsheet. 55

V. RECTISOL PROCESS SIMULATION RESULTS

The process simulation provides a stream-by-stream heat and material balance which is shown in the simulation output in Appendix A.

The predicted composition of the desulfurized gas is compared with the composition from the CONOCO Design Basis in

Table 4. The comparison shows that the ASPEN model predicts a higher yield of each gas constituent. This discrepancy stems from the fact that the methanol circulation rate was established based on the fresh feed gas composition which neglected the recycled hydrocarbon gas, as was mentioned earlier.

Therefore, the degree of absorption predicted by the ASPEN model is lower than that accomplished by the CONOCO design.

To improve the agreement between the simulation and the design, the methanol circulation rate in the simulation could be increased.

The acid gas and the desulfurized gas are related by material balance since virtually all the gases are recovered in these two streams. Therefore, the lower degree of absorption shown by the ASPEN model leads to a lower acid gas recovery as compared to the CONOCO design. Table 5 presents the acid gas composition for both the design and the simulation.

The composition of the regenerated methanol calculated by the rigorous simulation of the hot regeneration column is shown in Table 6. The model predicts that it is possible to regenerate the methanol to 99.8 percent purity, contaminated only by 0.2 percent water and 2 ppm hydrogen sulfide. 56

Table 4 DESULFURIZED GAS COMPOSITION

Component ASPEN Model CONOCO Design lbmol/hr lbmol/hr

Hydrogen 1909.4 1907.7 Carbon Monoxide 4048.7 4037.4 Carbon Dioxide 173.4 159.2 Methane - Butane 499.6 485.9 Hydrogen Sulfide 0.0 0.0 Nitrogen 41.6 41.6 Water o0.0 0.0 Naphtha o0.o 0.0 Methanol 0.1 0.0

Table 5 ACID GAS COMPOSITION

Component ASPEN Model CONOCO Design lbmol/hr lbmol/hr

Hydrogen 0o.3 2.0 Carbon Monoxide 4.2 15.5 Carbon Dioxide 125.6 139.9 Methane - Butane 2.9 16.6 Nitrogen 0.0 0.1 Hydrogen Sulfide & Carbonyl Sulfide 13409 138.7 Water 8 0 5.6 Naphtha 1.8 0.0 Methanol o0.o 0.0

Table 6

REGENERATED METHANOL

Component ASPEN Model lbmol/hr Hydrogen Sulfide 0.13 Water 8.84 Methanol 3817.90 57

Table 7 shows the heating and cooling duties calculated by the process simulation. The heat balance predicts that the refrigeration duty to cool the recirculating methanol is

10.86 MMBtu/hr. The total predicted cooling duty, including both refrigeration and cooling water duty, is 16.80 MMBtu/hr.

This is lower than the duty quoted for the CONOCO design.

The deviation may be attributed to the differences in the design basis for the simulation and the actual CONOCO design.

In order for the simulation to match the degree of carbon dioxide absorption in the CONOCO design, a higher methanol circulation rate or a lower absorption feed gas temperature is required. Both of these design changes would raise the refrigeration duty. As Table 7 demonstrates, the calculated heating duty is significantly higher than that for the CONOCO design. The heating duty represents the steam heat required to reboil the azeotrope column, the methanol-water column and the hot regeneration column. As such, the heating duty is highly sensitive to the specific column design. 58

Table 7

COOLING AND HEATING DUTIES FOR THE RECTISOL PROCESS

ASPEN Model CONOCO Design MMBtu/hr MMBtu/hr

Refrigeration and Cooling Water Duty

Solvent Refrigeration 10o86 Hot Regeneration Column Condenser 4.31 Azeotrope Column Condenser 1.63

16.80 18.76 Steam Heating Duty

Azeotrope Column Reboiler 4.22 Methanol-Water Column Reboiler 10.75 Hot Regeneration Column Reboiler 10.10

25.07 11.96 59

VI. CONCLUSION

The process model described in this report represents a

viable Rectisol process, based on the CONOCO design. The model demonstrates that the synthesis gas may be desulfurized

and the methanol solvent may be regenerated with reasonable

operating conditions.

It is difficult to assess the validity of the computer

model in view of the lack of published process data. One

point of comparison lies in the predicted compositions for

the desulfurized gas and acid gas. The current basis for the

methanol circulation rate, which is not consistent with the

CONOCO basis, renders this comparison meaningless. Unfortu-

nately, this project was concluded, due to lack of computer

funds, before a valid comparison could be made at an increased methanol circulation rate. However, a minor adjustment of the main wash solvent rate to approximately 3400 lbmol/hr

from the current 3100 lbmol/hr is expected to bring the model predictions in line with the design basis.

The Rectisol computer model was developed as a flexible

tool which may be applied in future design work or sensitivity

studies. The effect of a design modification to one of the major process units may be examined using the rigorous stand- alone model. The results of this simulation may then be trans-

ferred to the overall simulation to assess the impact of the

change on the process as a whole. In addition, the model may be employed to consider the trade-offs between such variables as the absorption temperature, the degree of regeneration, and the methanol circulation rate. An enhancement of the 60

Rectisol model by linking it to the ASPEN Cost Estimation

Subsystem and Economic Evaluation Subsystem would permit these design changes to be evaluated from an economic standpoint. LITERATURE CITED

Boston, J. F., Energy Laboratory and Chemical Engineering Department, M.I.T., Cambridge, MA, Personal communication.

Boston, J. F., P. M. Mathias, "Phase Equilibrium in a Third Generation Process Simulator," Presented at Second Inter- national Conference on Phase Equilibria and Fluid Prop- erties in the Chemical Process Industries, West Berlin, March, 1980.

Britt, H. I., R. H. Luecke, Technometrics, 15, 233 (1973).

Clarke, E. C. W., D. N. Glew, Canadian J. Chem., 49, 694 (1971).

CONOCO, Inc., "Phase I: The Pipeline Gas Demonstration Plant, Design and Evaluation of Commercial Plant, Vol. 2, Process Design and Preliminary Project Engineering Design", FE-2542-28, V2, 1980.

, "Phase I: The Pipeline Gas Demonstration Plant, Quarterly Technical Progress Report for Period 1 Oct.,1979 to 31 Dec., 1979," FE-2542-26, 1980a.

Cook, M. W., D. N. Hanson. B. J. Alder, J. Chem. Phys., 26 (4), 748 (1957).

Dechema Chemistry Data Series, "Vapor-Liquid Equilibrium Data Collection," Published by Dechema, Deutsche Gesellschaft fur Chemisches Apparateweses, Eds: D. Bahrens, R. Eckermann, (1977).

Donnelly, H. G., D. L Katz, Ind. Eng. Chem., 46 (3), 514 (1954).

Evans, F D., R. Battino, J. Chem. Thermo., 3 (6), 753 (1971).

Gerrard, W., J. Appl. Chem. Biotechnol., 22, 623 (1972).

Gmehling, J. D., D. Lui, J. M. Prausnitz, Chem. Eng. Sci., 34, 951 (1979).

Graboski, M. S., T. E. Daubert, Ind. Eng. Chem. Process Des. Dev. 17 (4), 448 (1978).

Herbert, V. W., Erdol und Kohle, 9, 77 (1956).

Hochgesand, G., Ind. Eng. Chem., 62 (7), 37 (1970).

Hoogendoorn, J. C., J M. Salomon, Brit. Chem. Eng., 2, 238 (1957).

Houghton, G., A. M. McLean, P. D. Ritchie, Chem. Eng. Sci., 6, 132 (1957). "International Critical Tables of Numerical Data," National Research Council, New York: McGraw-Hill, 1933.

Kirk, B. S., W. T. Ziegler, Adv. Cry. Eng., 10, 160 (1965).

Kohl, A. L., F. C. Riesenfeld, "Gas Purification," Houston: Gulf Publishing Co., 1979.

Kretschmer, C. B., J. Nowakowska, R. Wiebe, Ind. Eng. Chem., 3-, 506 (1946). Maddox, R. N., "Gas and Liquid Sweetening," John M. Campbell, 1977. Mathias, P. M., "A Versatile Phase Equilibrium Equation of State," submitted for publication, 1981.

Nonhebel, G., "Gas Purification Processes for Air Pollution Control," London: Newnes-Butterworth, 1972.

O'Connell, J. P., "Some Aspects of Henry's Constants and Un- symmetric Convention of Activity Coefficients," Phase Equilibria and Fluid Properties in the Chemical Industry, ACS Symposium Series, 60, (1977).

Perry, R. H., C. H. Chilton, "Chemical Engineer's Handbook," Fifth Edition, New York: McGraw-Hill, 1973.

Prausnitz, J. M., "Computer Calculations for Multicomponent Vapor-Liquid and Liquid-Liquid Equilibria," Englewood Cliffs, New Jersey: Prentice-Hall, Inc., 1980.

Rousseau, R. W., J. N. Matange, J. K. Ferrell, A.I.Ch.E. J., 27 (2), 605 (1981). Russell, W., Badger America, Inc., Cambridge, MA, Course notes, 1981.

Sander, W., J. Chem. Soc., 121, 2667 (1922). Scholz, W. H., Adv. Cry. Eng., 15, 406 (1969).

Siedell, A., "Solubilities of Inorganic and Organic Compounds," New York: Van Nostrand, 1952. Sobocinski, D. P., F. Kurata, A.I.Ch.E. J., , 545 (1959).

Spano, J. ., C. K. Heck, P. L. Barrick, J. Chem. Eng. Data, 13 (2), 169 (1968).

Stephen, H., T. Stephen, "Solubilities of Inorganic and Organic Compounds," Oxford: Pergamom Press, 1963.

Toyama, A., P. Chappelear, T. Leland, R. Kobayashi, Adv. Cry. Eng., 7, 125 (1961). Treybal, R. E., "Mass Transfer Operations," Second Edition, New York: McGraw-Hill, 1968.

Wiebe, R., V. L. Gaddy, Am. Chem. Soc. J., 6, 76 (1934).

Whiting, W. B., J. M Prausnitz, "Equations of State for Strongly Nonideal Fluid Mixtures. Application of the Local Composition Concept," Presented at the AIChE Spring National Meeting, Houston, 1981.

Yorizane, M., S. Sadamoto, H. Masuoka, Y. Eto, Kogyo Kagaku Zasshi, 72, 2174 (1969).

, S. Yoshimura, H. Masuoka, Kagaku Kogaku, 34, 953 (1970). BIBLIOGRAPHY

Rectisol Process

Herbert, V. W., Erdol und Kohle, , 77 (1956).

Hochgesand, G., "Rectisol and Purisol," Ind. Eng. Chem., 62 (7), 37 (1970). Hoogendoorn, J. C., J. M. Salomon, "Sasol: World's Largest Oil-from-Coal Plant," Brit. Chem. Eng., 2, 238 (1957).

Linde, W., G. Ranke, H. Jungfer, "Ammonia," Fertilizer Science and Technology Series, New York: Marcel Dekker, vol. 2, 233, 1973.

Maddox, R. N.,"Gas and Liquid Sweetening," John M. Campbell, 1977.

Partridge, L. J., "Production of Ammonia Synthesis Gas by Purification and Shift Conversion of Gas Produced from Coal," The Chemical Engineer, 5, 88 (1979).

Scholz, W. H., "Rectisol: A Low-Temperature Scrubbing Process for Gas Purification," Adv. Cry. Eng., 15, 406 (1969). Acid Gas Absorption

Kohl, A. L., F. C. Riesenfeld, "Gas Purification," Houston: Gulf Publishing Company, 1979.

Treybal, R. E.,"Mass Transfer Operations," Second Edition, New York: McGraw-Hill, 1968.

Physical Properties Models

Davis, J. J., R. I. Kermode, "Wilson Parameters for the System: H2,N2,CO,C02,CH4,H2 S,CH3 0H and H20," Fuel Processing Technology, 1, 247 (1977/1978).

Mathias, P. M., "A Versatile Phase Equilibrium Equation of State," Submitted for publication, 1981.

Prausnitz, J. M., "Computer Calculations for Multicomponent Vapor-Liquid and Liquid-Liquid Equilibria," Englewood Cliffs, New Jersey: Prentice-Hall, Inc., 1980. Prausnitz, J. M., "Molecular Thermodynamics of Fluid-Phase Equilibria-" Englewood Cliffs, New Jersey: Prentice- Hall, Inc., 1969. Reid, FR. C., J. M. Prausnitz, T. K. Sherwood, "The Properties of Gases and Liquids," Third Edition, New York: McGraw- Hill, 1977.

Soave, G., "Equilbrium Constants from a Modified Redlich-Kwong Equation-of-State," Chem. Eng. Sci., 27, 1197 (1972). APPENDIX A

Overall Rectisol Model Z -Jz0 0 -J -j OzI W 0 0 I I. U) fz in IL - IN ~Z U) 0 0 0 3 OZ srz IN (A ft U) 0 ft 0 Lo IU) S a. 3 U) Z- t m a -j (in3 cpnoft U. 0 O(IL a 4 -Aft Ifl-.1L -i U - N M 0 IL .u a 0 U. U) U) Clta 1 I LLL. . - X .NU. U 4L. I x U) OY)u4 0 ) 311 - W ) U)Cz0 U) 00000a 3 i 3 C z U7 U ftZ I H n n4HU ft 3 . Z.2 9Kz C Of -jI S.0000~ S I Si3 a. IL CL *e m 1 ft ft > *i CZ U x o3#INM 3 o o 30 - II 11 a U 11 1 333330a0a U H 11 11 z * cf * t .. .J U II I I.-I H I I II JOtu ' * LU * 9-9-9-,-'- * m * 0 000 00000 00 00 0 04 z * ft * ZLU-A U) Z * 0 * -j3000F * Uo * 0 O. * m * CA > -J * 4 * U) -A - UL.4 4 * * 30 0 a0 -1zI I (n 2 II I lz U) 3 0 x * U * ft * z * U - U * 4 * * 0 * a. IN aox fe 00L IL 4 U zS * P * _ L L. - X N - Z Z X -C > U * I..3 * U) - nU) - U)) *aJ w 0 SOI IN * 1- * W n lzf I'I0U 0 x x- O al LL3 3 l * U * 3 3 U) CZ z x N oa LUX Z * 2 * 3 3 3 * LU * CZm - n * cN 4 aE _f U 0 * LUZOLL A a U.U.0 * z * #* Uz *m* a a 0. 4U. j U L U IL 0. 0. '1. lL U) * C0L * *5 xM~ ..J - -J J J > : 0ftLU"x Z U. * 2 * N 0 tY * 0 * z I- ftLU U * II II # * * N U If 1 11 u 11 U N N n 1 H * P MIMMIa- - a- u * O._- * Cl 2221 11 I * 4 * 22222 ZZLZUZZ * * Z * * _ H _II I * 1- * ca* o * * * m 1 U) * * 4 * Z ft u 0 4 ft * Oj * f *N4 1- 1-a. * lY * wLU tI X ao LU I O-i * 2 * L I X * CD n01- O * -J 0 * z * ui tI - _ Z Z_ 4 J u00moa. U x * 0 * * 0 SIN .> J S * LU * ft :EMMMMUMUs 0 :] - _j x' .)0 L_WC , < W * * a * I a.U f 29 * -'0 4 * L * UV) Nj 0 zuY * Qx 0a. 20 * U) CU U)U U)U)W *a X . S L U I 1U IU U4LU v - Lu * CY * -9 0 1 4 a* 4 *U)CI * 0CYU)..IL Iaaa C 4CDCD' N Lo0 w cU)0 0 * 42zU U) 4 * NfiI I I S LULU * J o .2 1 u . . I M M 0 * mOO0OI-I * *4 LU L U L U U) U 4..J 9 -2 * I * LL. J U)09 -. * SO. 1x * C>I * U * q" m O m 4~ ~ mmmmmmmmmmNNLU OU * U) * YYYYYYXYYY n144 LU 4 * * -C * uuuuuuuuu 0 ua * .J * LU 00O 20 uo * 0O ZO00- 4LU00000000001 U U U U U 000000000 1a co0c f * L. * w -j ------2- O ftU0 Z U}1 U I IS Z 0 I ,mmmmmmmmm = 0 tZU= 0 4 U) a. I ...... 4 4 0 0 LU31L - A..J- - - ...J...J...1...J. l 3 o Axtni 0 0 cM CL Z U. .~~~~U...... U. †. OI- LU cx .1. NO X CZ LU4 4jZ - cx 0 uZ O I X CD 003 -. 0 (3 '4 a r'O00 0 (N I- Zw]i0 CN I^J N CN I O. U- Ii044N LAp- ON C: U LU ~XlIZ1N1 I Oj 0 N 0 11N LULL N 1 I In L-1 p- x I>- 0 II II 0crx O < N - u-- (.O0 LUJ '-i- 31 3 14>xc n, 2 II It ii Ii ii ii 1ii I Ii i 11 1 In 3z, I, LL F F -I- 1O 3 3>0Z M m -- 0I---- O0 0 300 o o 000000 a ZOOa 0 -J U. 3 N a! O * U * 4 0 a. * 0J * U1 CY LJ - a LL 10 (O O * Z * 0 ~30JIo > 0 U. I * P- * (N U) O W * Z * cww00x LU> * Z * 0 0 * : * C o * LUQ * ~-(3 (3 ~O * 2 .* * * ob 1IIII * -j * U < N NN * :3 * L.1 * J~ * -. N U.Puuj * 0J * Cl z cjx~xZ~0LL4N1 * * * * * 0 * O L * tr * w 1 ii * O * * U * II II II II II * Z * * O * NZZZ II II ZZ Zz * Z * P- * Z 11- * c * zZ 4 * * Q * i=, ZZZ * I4 * II II CII * D * () * -Z 4 LU * 8 * U * I- * 4 * .40. _ * CZ * * O * * Zz,, I--I- * -a * 3 * U. * * I. * LU * * 0 * * cx * cx Z * O * 4 I 4 - 2 P- * s * * * P-* Px-~ * P- *. -a * 4 * Z -J * * n, * U C * U U 00I a- * * * L * XLLf * * O--Z ~-I-i- * 3 * 00 Cr330 0 ,* 0 * LU * LU * LU * * 0 * O 0 * I i -LU-_ai 4 ZL 4 0(.- fa (9 -a Y Y Y * V) * * * * Z * . 0.r (3 * .d * * * LU (3O (3Oi n, * * I * ZZZ N4[4[ * 4 * P- * Z * - * 4 11 , * * * r * * c * I * Z * (3(3(30'LU * a I * YY * O * Yv * , YYY N o Cx000Ilm mn,, * U . * I-0OO)0 * "' * I-(3 (3 * I- * *4 * -003( 4 LU * -aUa-a * c * * N * WOO * ULU * LU- * o * UJoo o Z * J J.J C:[ 0 LU-J00 .=-- *-.- -mJ 4 LU * *LU (0(0(00O0 =JJ 2 * * * tJ 49 -rI m mm * I * LU..4..OM coCI * Qx * wS * 3 3 3 U) 0 3 -J -J 0 o3 ...... ~~~~~- . E- JL .... i0 ad 0 C cl Z U)3 z

0 3 0NCl 0Ito > 1U 0 C. UXr U ILU CUa Nr >30IzL Co N000 0 N CIDC.) ii'I I~~~~~I M Cl 0 0~~~~ I it L -J O _04 N It U- . . IO II II IA I q t II 0) III) C1l Z'' I . IL O OUIIt In, In , LUn · 0. qY fULY. 0 1 o * 1 0 .I- ' #Cl I'0 -I n 11Lsn znLU LUH ~~~n -0. lz * Y IVo -- 11 L 6 InWr LUJ uj LU O w- u · a 0. H ' 0 33 U 0. 0.C DeI 00Z I ILz~ Ii * U > 3 J ·~Z-* Z-~. *# wr0(n..0C* ·0~4 2 4 Z tD * u..*~OZZ~ ~ * F N ~~04 tL C.O q ' Y OU ill N Eli1 . 0 *. Cl * O * .0)O cz I- * 0' 0. 0.' 0.-' LUCi')t- 4 * * NI. LU Z Z Z n4o Un o inu a t. * * 0 N U) I z CD0~rX s~~~~~~~~~1 11LUNtlO fi s rzCq l:I IS CLJ - * t : r c Ic CX LI aX -* * t 4ClSl x LU' x X zn LUU wI LUc LUo C * 0 4 UN -U) 0.IU P. F 11 4- 30. 1- 1 1 . * * * _ * XU) L l U , U)0 Z4n *s 00 3 3 3 3 .j3 * z *W *UC * in CD I- LUW IIa 3 CN 4 0L 4 U-oI- 4* I. LU 4 LUIZ I 0 O0 -0 OO .o * - * 0 * az r . . N 3 II 4 4 4 Z 4*. u 0-i CL-. .. I --I * CZ * *. 0 U) U) LU. IL' CZ L LULL I U. * 4 * U * C - c1 rz 4 z * *LU Z LL. .J · 0 - N 0' 4 0 *Z JZ Un LUr0 Cl Y. 1- 1.2 uJ 0W LU W LW 0 4 4 in * X 04 - . n U) o 04 LU;, O0O ( - *..J 4 4 .. 4 2 0 Z 0iJ 0J J 0J 0 . * o a u , Im * O U . * O 0 * 4* * * 4 4 oIn · ·· h-1 * IA0 * * ZO * I u.. J IA 3 i~ *e e. ~e * Ie le* e e *Ie f~ 4Y -4cY 49z, 4z 4 C CO oco 0 0 0 LU LU LU LU LU 0 0U 0 -J *- 1- 1- !- *- .- - - 00*i -i IL .- If] .In~ *.-l *.-n *. *..0)...... [ .-~.- .mco .. c I I 0 C, xCl Ca IZ O0 0 0 (DO tD. I° 10( 0 00)(N C ::O -N IN 'rmcI, -"IxZU (N4 0. 0. N - 0. CNd0

IC _H'11i Y00IIe 0. ZZea- 0. Z N- _ _ 0-CD CD z #t 'r' Om Z _ t U 00 C N O 0 0. CO Z Z 4 (N y u. u(N u V)xI N*I ~.~ # t)O OD v11 11 ' O. (S 0N.a {3 C)0. z) 0.(')Uf- 0 In.

n, >. . (/)If)V)I 0]N .0 0.~ _ ut/OU C')IcrIV) ) m / ,.~m/ m -v 3OCZ r 0.11II 0.a' WL D 0 It z U. w-0 if In ZuZi I C I11 * * U 11 H uw C'CZ Ma U) (' * z * > L. > .W U- .U- U 0 * 0 * .- 000 · 0) * - .- .. W. '0' V33)(lf.' 0{t 0. I-* I-ti In ( NO I.- 0 .N . c0 (N U)co o _O 4 O * I (N o (N I.C W N . .. .W 0 ix: , < I-."uL coN J 0.N 4* ( I 1 m * C * I M X 0. > i X c 0 I:Y 0.> Ce 0 Z:H * W * 4 0 0 Ct U' U)U1 X V)O * W * 4' - W I.u 4 UJ V).J W U 4 Z * .J Z W I 4t I-. 4C ' * m * J :EU Z GO O U 0 co0 -11 IL X N ILL o. * W * If = 1 F- 1l J Zs it * t * LL cL Uu O I U if a) 4 Un _ 11 I- I- 4* c * U 0.1- 4U ) ILV) W IL V * 0 * Z- W EA I) 0 WAJ *-~ I-W 0. 4 W 4 ) V)X m cc 1W = WWLL W )< W * ) * W I 10. CD 1W -..V) (.y -ZF O * I : 0.0 I - ZX ( N14X * 4C * (flu- (/)4 (I.l.l) 4I U0U )N 0 CDO 0. ) co < - XZ In00.. S .4* I 4* IL.J0 z . 0. C: . C0t ( 0. .-. * ILZ 3 31r l i-I > 1 WUJ WZ 4 * V)z 0.1 I- II I II U) * W4zs W WV) a of 4 -.10J -4 10 : 4 4 4 4 3 < x44 V)4 V V)4 (N * O! tY tY 44< -j II u WJ o~04 o.) 04 Y*J 4z .. 4 crc W c W~tY J~ .J .J c UIx14 s iVo:W u ' UC 0 0 0 *13'0 11 U- IIq~ I4 II tO

11 UJ~~~~~~~~~~~~111A .II 0 ~QI 11 II N 0C >, U1...I '- C] wws'00 0.V) 04YY", H- 03. ,, 0 0 t . 0 4 CZ> 0 .L0.0 I 0II ' C0 InC 3t I . . ~~ LI~Li~ i. * Z* 002 ' oo 0 Z ~~~~~~C)NI 'I 0 I' II W '-C .',> 2 cR u 4 NU. . 0 Yf.Q.Nf 0 H-C *0 03 I I- 11 UJ If _ * Z * 22 - 010 0 ZWW 0 0. 0 w NA 11 0. 11 CZ C zZ0O ix O LU~~~L 2 o o II *II 03-ct I- - x NIIIIII II *. , ~*'T'~0 tuj II c 0u * W ~* 0 * Wz * Z * I U) - *.O C) * ( I4 * 0 - 2 ~~ ~~~~~~~~~~~~~~~~~~~~Z 4 in Z I V) * C 0 C /u) t- N 0'20 -. I C x * * 0. * : * II * 4 * C NX3> '. ~ X 0. U. 1' W * * 4 * " . 0. * .. * N W U) ONNCV)< rN U.LL4 U (1 C.)3 20 H-OQN *n,CX CX I - Z2 0 O 4 0 .. X > xI x * Z~ W * .10I : 20 3 3 U O JDwN U0.U W1 U o W in * 4 U) . I'I *> * I * 0.0 3 XI 0.11 L. 11 I I * XW * X I4 i 0. I- I * * 4 ( U) c. n z U. 4CIIC W 11 * JI * U '2- OW 0C.)0 W * * Z Lz Q0 WW 0. * W U ) 3 * .~~~~~Lw .*~ U[.ww .t~ C) U3 * *C 4 m 0. 00 ) ** 0 I * UJ X XU) I 1E0 0 20 Z t C4' Q' * ,I * Z O0 I Q * 00 * 10. CD 0 * 4 * 4 0. L) LI.3 0 . tn m 0. 0 0. 0.Q 0. 4 I * 0. 0 . 0. * CZ 4* U = -J * o * - W Im 0. z , * W : 0 *4 * J v) L. * Z- * 2 U) 0 W ( u 2 O 0 C **- I * . o x Ix > -4 04 I.0.U4 4C 4 44 0. 4 4 4 W CAPU0 Wl rQ- . OOCX ,O,3 O CX (3J WCX O *4 4 *4 W Il J WX Q Q QxX rJJ Q * 4 N *4 N 4 ..I N 2f W CZ< * * 3 3 4 .... (3'.X CZ4 M a4 m * Z * Z Z < Z- I 4 -1- CZ < 0 0 * 2 * :E 2-J CY 0 0. 0. 0 0. 0. 0. 0. LL L. W0IL(3I U- L Y Y Y Y Y Y Y Y Y 4J C Y 0 0U 0U 0U) 0U 0U 0 C) 0U 00U 0 0) .0 * m *m *m..(D.c *.(D *D** .10*-rlm...... com *(m *m ..J ...... -m I I O 0 00

0X cv0 '' CD an f I ~~~~~~~(D (DOI0 0 . (N. 10 0 In C-4 _I C4inO N1 0 I r4N I c"IN 0. i Z 04 I _ NlI in Y U I0. x X

0. CN0 _11 y0 ZY 0 1 0. 0. 0 v0.C. Z O 0.Z 0 t Z C. Z 0 Cl LO ID U')11 CII Nq-) en N N in in N I I~~~~~~~~~~~~~~~~ 11I00O 0. 0. I t I 0.. 1cn In 0. C)0J .fC ,.0 0 0. 0 ~~~~~~~~~0in.0 * Z ·* I 0. 0. '-' * * *"' C~ E C~ , ,,u~ * * * 0 * IC CLI- Ow 0Ur LJ W4 r.- )~ ~ . , FI--I NI * * * i1 1N 4* in0 4* * 0 * X 05* * '~i * ON 0 U0.0.on JXN *Ii * Z0.J * W W, * AEZ * CD4 IinU. 4 Z n0 * "' -*s 4 * J- * . *-J - Cs - * 0Z **.> . '" * 0 *-. J oU30 wX * CL * 03W 40* .IX* t O J I 4 4 * * 4* 0 O 0.* .IX w4 P Z N *4 I * I I 1 In * C * -J x .J * _J~0. * O4 * * ~wJ ~~u* J " --. :In 1 I- -n W- * * W WI 4UJ' * . * - (N _ o0I * 0O~ * -~ FCYn ,0. n 1O X 4 11 LL N 11 * I 11 it WN * I * w 0L i 2 _ wU . O0 . (A tO In * In * WC) N 0.1 4 W * 4.* W 4* .WJWn.. - .. CoiW in 11 [-- * * WW l W L W * 4 InW in *,* "r CO * .q-.I~ 4X q-I.0. IXWin lY JO * l * I I t * 3 * 0. CD 0.0 4 0 * * W 4 0. E WinU} X 0.0. 0.J-J > * * 0. 0. a0. * * I : D0 * F * 0. ,,) u- .L 4 * < * 2 J UJN0C( L.. :E Ou * C) * D In _ * Xx in in * I- *4 (D1.- sb1 0 cC CDI W I 2 X4 Y * > * -J 2 -J Z 4 W 4 - 31: 0.s * 0 * I- * 44 1-F i cUY) i 0. ..' 4 C.'* O4 O 4 IXL * 1 On0 U 04 4 W * 4 . 4 4-JW J ... U co IJ..L-J 4 W 4 o r 0 W * 4 4 3 CN 4 (NS 4 N 4 44 O0Y 4 *t4 UU U*- * 3 * 3 I- i I4 4r* 4 0. 0. 0 0. LL u. 0. *W Im m- C. C. U U U U) C. U U 0 0 0 0 0 0 0 0 0O ... I_Pm ..m...... m O ...... * m.*mCD .=. o ...... -* .m ~DCo *0n *e* 'r 3 L-i3 -J0

Li11 -J o

> I M 4 w 0I 0 0 Iu 0 > C', I :II (DI 3 11 Z 11 z F tn _i ILU I LU IU zz-J zLU 'Z Z 0 -J 0G 0 0 zc0u.,'r LU0. 0.L 0200 0.LU a 0 O0 - I X O 3· uY 0 OZZ> U I-N -j JL F IN 0 * * W I L. O LUO Z 4 L)U * * L L WUWL 11 I- * * I I -ll LJJ IN LA ( .- I I * * 0J L< . 0 -J * * zrN -J tr N I N J * * . LUI I .. 11 i 4 11 0 z r 0 * W Xj * 11 -J Y ,,. hII LU (.10 l (*y I 0 u :E r u * 0 * Lu I.0 uJ411 a0 * J * 0 C :r 4 . x * - * t Ix 4 -0 * L. * - u I-F IJ * F * F - LU X 1 4 Lu 4 W .J X * * 4 * {3) * Qf 4 * *4 4 * A.4 C/LlO LU * D * 3 * Z * C 31 I- 3 -JO E * CJ * 0 I'.- -- I * .J * 1t * 3 0 3 O Fe u. *4. .X + U. 0 J 0 N 110 * -J * u.,- LJ u. LZ 4 N OI, * 0 , LU0) -t X) u. u < IJ30-J40 0 L IL Qa,. Xs I. .CZt U. 3:- * 0 * J O Y II 0)3 --J ILUIe>I-( 4 .I UfUJj 0 * * - 0 .La. Lu.J u I * t ut z ct z * Q 4 JLO .J Y ZU-1 LU 4 4 4 LU I J- Lu * 3LAJO-OLfL . > * OZOUZ0WF-M * 4 * u. I UJ 03 ' 0003 > * I. * IA * et * 0 o I .J..J C S u 4 3 4 NcI u., * I * u. * ~ * Z 3 --- X CMeu0=I~ LUfCD Z cu -o LL0 ulJS <- l F- * Z * . SJJu.-j cn D * 0 * 4 WI .J CDLUj ) uL 3 X -L fL 4-0 4a.. o , -I 01 LU * Z * 3 a. II N * F 4* Q '3ZI. Z 0 3 -u._ 4 4 * -J 311- U 1w1LX Uji UJUj e- *uJ * 0 * I * 'r-JI- N Z _ : * I- LL 4 itr . La.La ca. er i-'-4 -11LU *s * (.tI * Lu 4 IC') . * .- U. X * .J uJ LL * :E * LU * * u0 L L L > . a. ZZu; .LQ*u u13 u. * * z 000Wu O CZ --0 -' LU CZ 'X

..I-C) ,L u. u. * LL...... F.r LL.. LUI .4 0.

OWO(D

LU I- 04 ,.,o .0. .. *.. (LU... Z Z., 0 -j-l * j I n 0 . q - z 3 . 1- *W. 0 *InV)nWn(A W 9L) s(.M -, 2 .0 ,C D· 5 O0. ,3 2In22 I U)n Cn0. m U U CZM IL)02--L u 0. XW X 9 unID U .4 2-jIL 3 3.4 41 > ZW WA .1-III,~ 0 Ocz L2J-- N W 4002 *InO0 ~0LU LUWVnV.4 -II- Z 3X InWn U CZ)-OUL 4)XX 10 1U 1 1U2 4 *i-IO0LU Z, W.C.C . 000 UUWZW JLJ j I - U0WLL U .4 LUNN3 .00 .L Z O W .0a. .. . .1InIWmm Mz XILLIL L Xcc zZ z 14 42z lILCOU IL- >- >- LU -- Z LU .4 z.: LUa 0 CZ~ mCZ i i- C ~Ci Cd (N a l m m m I m mW *M LI M LWUXC m W -m Lu X ICZ mI m u M l LUW m LULJWam mLU WUW U. -Iv 00 *I LLI LU UJ ..J LU s UJ LU In s- In M J WLU -J s- - -J 0 WU - M LU WUWU WLLU LU WL s- W In W LU 4W 24U WI.,..I WAI-JW ..JLnI,- 0 .431 W. 1 1..J 1~~ ~~LUWU j I- J -i - Z I -i I.- F-- W I-I0- zz m I,,-,-II. I- 4 Ci 41- (N Ct 4I4 I- 41C- I - 4 I-- . C C C 4 4 Cl . 1- 4 .4 Zoo 0 -i ~in- ,0 WI W I 'WILU LU LWII I W I W I-LU W U) - LU X ix C0 of mxs-s 0 10 0 OM I 0 0 'm'I, Il -. 10 I-- In 50 I I 0 0 1-i .J I 5 I I Im l I 100 0)I I 0 IJi l LUWI. mOOWW . I.- .j C I 1- Ji0. 0. Cl .Ji (N4l ( I- .Ji C('3 J- I-- I- ..J..J I.- .J 0..-j 3- 00 ~ 4 .. In .4 .. 4 d 4 4 4 4 4 t W WLUCaL W WL ..4 v UmmmmmlnmlmmmMI m, mI :mmm mMmMMmmmmmmMM In In0LU W 0. W LU L 0 > 49.49.C44nU) . 4 .4 .4In).4.W IILLL OO U0 I 0In Wt uWU 1W- 1W3 z 0 0 U. 0. Z 0 In

0. In .4 LU C, 4 a.

VMW(oW 00 ON V OC- LflM M L() Dfl,) l w C Dwww 0w w C h - I ~~r~

. . ..

... .> ...... 3. .j . a . .. o. ... ,,...... ,.N..o.o..Zo,*Z. .. o Io.o ...... U > o mr* a . U 'Z ..N..W I.. :K o0 M 4 x -3 1 .W .... -it ...X>0 . 3a.jX ufj N a .> . -r. C U, Ns tn U. 7 o o 1 u 3 N N ) U 0 mtflO a xN > CN .44 * >z3u> -0 X4aj C z l '3 I tL CZ;V )4 J L CZO LQ 0 > I i -J Z IfO Vou- 0 L) UtX>JU< 8 Q1 >3 ui 0 I-. Z00(A0tn ClO*I O*03m0aNX s 0J IL ~ZZCN-4 I 1NZ. za .O z 03 . /)a.OfCCZx 1-3u-I -J 4 JCY 0 O- CY' C3 0 m L X N.Z 0. x a 00 LO LENu-S OX34CIU>- > MQN0OI C42I. 0044JO C3 uj I- *I.CWU.2J- >0-< Z Z0 0u'3-MOma czW0 - I.. tz a X L O CM <0 1X CY L CN U XwJN 01 M/ C 1N 3 X 4 4V .CO-wel- ugx* 0 u LL. : * L< 0 C wwja -0 .. ;I0 UlL 0. X 40U .) a a 3N wwu Zxa. fi-lL-4' OI-4 O.'L Z I IL O J -JZr0 U " UAJ w*I- )311=c Lt Yw Iz ot wei1 2- wu a.= 4 0 0 II 0L > U.- - > Z> 3 ICZ N -

zuLJ 0 z 0w In

z ILa. In 41 .

CL a. a. I- -:E I 0. z N z c I 0. V) M O x LI X 0 o0. 1-4 x NI I : )UccI Lfl 0 Q 0.1 - I M)= 'n (n U) Ii.-j U) 1-C4 x (n 0. 1 . C c I Ia. wo t wcx)= X U 0.I o I LU1- U U)U)I 4Lu1ax - Lu44X >. V) . LL I 0 =N uAJ x l- 1 COW 0 0 0 C C -J N4 Z LwL ICJ I - OUJ 3Z40 L C 3 Li C) I uLJU. U CI CZ xE Z ~ Ix U) C.) c I L I C- z z wIX(fl- 4 03W CD 1-1- 0. 4 0 I LuJ Lu Lu WLu I U 2 LU . 00 (nin 0. 0 z W in Nz E M C.) 0 0 - N UJ CY UV) 0. in C)<) CMCZa ') Cy w Ui lw U3 cxJ E Je U)tY 4 4 ->WW U wU. Xa.0: m xC-:43w. Lui-1I X ~00ot Ji Z o n z 0 3 COC.)0 O O 0 04 N4 1 0 0 incx -j U C) LL V w U LU V N s 3 )0 0 tnI U 0.0. I u CZ M 0. Iu u. LL C Of I z z c )N (nO~ZO Ir1- Ux LI IZ a: 0L NZ4 WoxI C .O -I 2 cU - -I z I I 4I .4I - 0 < IL l,-X U 4 IO O 400 oCxCZxaLAC< OJ 0 r Ui C) C Lu U. in CZ LU z LuCCY U U) Cxin 3 U) M4U I tr 3 3 r a UC) Ix 0 X- -IU.C.) I 3 cx co - LU 0. LU. 4N nnf cx cxw a. 4 I mxcX < N U) < N Vn33 I- -I tn tncx >> m xr w U L. U. LL a. -J J > W j a lY a > Z -j< 3 3 v J tn aU) 3 0 V) -I -I x 4 -N 4 n I I -j x I I 0I (0 V) n X I Lfl. U)>- a. 0 x 4 0. J - F- .C< _I < t- X u Z X 3 3 I1 3LU .do n - '- _n x @ JJx (n 0n I- In 0o-' UW uc U)CN a tY WuIu LU I- < Luz 3~.nJ Cj ('WI 4 0 in cZ w x r LuinCZx4 z u C 2a. co x Co u CZ U r o _ju03 3 Lu 0 - 3W cwI a 4j 0 3 C,Q 0 1 3 c.O t) L .co I CZm IA c 3 I: U 3 :L X ZI Ou~ 2 3I 0 3 0 -a N< 3 O I (AI 0 I cx I.-CZ t X 0 :E 0 . 0. Lu LJtn c, -CN I- ..I 0.0a N : I Lu II C J -J U. u U) a: '-CNI ZCx Z U 2 u 1W 0. cxc oU))U 0 .QXmQU J 2 Uo O O C) X 0. D O Z U) Q- I :-J UJ. O M OI Lu I- OWW Wn4-.II-2 0 3 0 X X CL 0z00 z z (Lu 0x IMcQ (UNLU O0O .) 3 3 C.) W4ui 0 z O IU-QC.) Lu Qa.a0IuxwmQ Lu 0. U ZZZ O Z C. III : 0 U) Z L u Lu U) U) U) OZ 44c (n in Z Q: - W .--- I- 0L.C Lu I Lu3r Fn N F- Lu LuL in -J X WU. Lul Lu z w a Lu w < inc ,.1 CNCN O. wQ ~ Lu U < 0 -IX Lu (NoJo U < . < Na I IW I < aJ u uL.C0 in I CWin (UnU- U) C C 0. wU)I tU 1= w a m QWW IW En incxxX mO Ln1- of ( U)33Luw (n C 3 3 a a c ov O : u. u1 3I W1n,,tu (n CZC axLU 4 3 4 cx in cx 3 x 3n J -N 3 3 F M -I 0 o31- inOQEx3Lu _ox u o (n a 0.WU = 0 )0.n 0 1 N 0 t 3 IV o W 1-IC,0 - u.1ouaffi Q J0. 0. i.- cx .j > ..Q-i > J U a -1 z . z -1 L Lu -1.U.J- >x LL Lu U. U. U. U. cN

LU 4 0.

i|X CD -J ()1i-CD' F- : 0 '.t :E 0o LU41aw IX I LU U I Z -j -J 0 0 L. L. CY I- o n CZ N -J LU IT UJ 0. cm In IInC 3 3 t 0 C. (<) I :E 3 Ix cr -J Z cn IX N 4 x0 0 UaQ x 0. L 0 CZ I--V 0. c > -A0. LI) -J -A . 0 V) Z -J LI U. 0.3 I-0 InLU -J -J0-VJ 4 - 0. I 0. 0 CZ . ; X LU V) O LU 000 Uj _ O 40O ~ '' . U LL U. lC U tr U Cvn C/HI=.' O[CIL3N U.V a IL In J~ In In I-- L. U. 0. . 0. O Oot 3 3 3 3 i-aLU L - Nx0 C. OLU z -J J w C . . Wz u ,__n, i' D 00. . LL-j -J> I I NZ O-CY LUC O 0. 0. U. LA 0U 0 0F U.L. > a00. 0. .j> LL .> -j ix of _^ N 2 n z 4 L) 4 < U-., LU i- LC1 0. I..L U a0 a LLDIn J w i-- x F- 3 In 0Or a I) 3 = 'r z F CZ 4-t .W LJI- 0 InIn a 03 0. LU 4-jj Q0. - X -J W - 0 I I.- -J -0. Oz 0 > N0 U) LL0 CYI FwOU0.- 0. In) -S -I LU i-n LU 0 LA. 3 I CZ In 0 00~W a Inm -A 0 LA IN3 0 2 co 3 In CO 0 x' O -jm 0 0. i- In Q : ) X 0LU- I -1j o > 0 co 0. In 3 U a0 2?- z U In3X .4 co CY I C In) > 4 C~U x < 0 -U. U.QQ mz'-X 0.InA I- Q 3N vf )).- ut f 0 C.) mm U0 - f 4-C I F- U} 0 C40 I LU U. L I n) ) - U7 U) 1- J - In n n ex Ca IU) LU *o a XI I- "l-x -J tn 3 C!Uw1 3 3 3 aX I z In' Z J x Z Z 3 CZ Ini-InXni-0 z Z Z -J -J>I ZLZ L X I 0 3' 0- X 00. ocr J J cU O ~ J. 0. 0. J U. U. U. '0 0s z z 0. X W Ln tY U. U. U. . Ul. I. 0. 0. 4> L. 40-" i-! 0' -- ..j - - - > Q Q: -J Z LL 0zLU Z lx I00E C., 3 'I z C.) UU J UJ ~0 0 0 I 0 ZZLU LU U . LU In In IIn I) LU t-X - i- I-- (N i- - 0. Q - 4 0. L..J LU zN LUJz LU LUJ LI -j - z LU X LU (N LU4<0 0 I Ix ,, ,, 'r wL4 LU U O 0 LU 344,"CNn CL X CL : LU Y X : U (Ny LI LI X U In7 ILU LU L (,u -I0 - n 0. 0. U w- -1 I~ J I Qf tY i ULJ In OX C4 I-xU U. !--R OU. 3 Ln 'nco cun > ZL UJ LU0 -j 2 In0 ' U00 v 3i- U.3 0 3c LU Luo3 *I I I 3 Ji X -0 3 U rr X z z - Q3m 0.X U: Ci > Uf L 0-300U. ", ") a03I 0 o 0o u i O 0 c Q CZQL N1 N1 Qi.i- -I-*L -U - C.)O UJ o cou. I Ca a -A 3 j < 2 -C 0. 0. LI Uf m Q 0 LA. U. UL. M CZ (O -j -j .I:I i->) C IX I [1.Z "r I mt0 Qf -J z LA. Ui. LI. LI. L 0 U.- U. W 0tq u0 0 4 000Iq I 00-0 C. 00000co t CDI LOLI LrI' co I 0O 04 V 000 00000a) C13 Ln 0m r-0 "It M ao ( n u(F (X;:Ta0 00oo-O (1) v 0 + D-0I +co0 0 am um 00 0 b3 1! t. 0000WO r. .4 6 0 (0) 0' V)00000 040) V ( O N N 0 C.] O - I. + 0 + + U, in 000+ + + z 0' x o ) ( C; U) z oq t N IV od0IT 0 3 - I-r- 00000+ + ++ + C) 0Q,0 0 NO U. 0Ix 3 Z Z CY '-0-0 -j ul Ln IT naa0 tn > J 1- W to U) r~ 000oN N( - tD Z F- 0 I- 0 N uj * O 0400)0.0 LO00 0 Il N Wa0a w - I:I r.- I NL-r- --N a) W0. ~-w M J N N I-*j L a Oi 0.Za ' F- +00001 + +4+ - F-W I I WO 'II (N w 4 J I -C I F- .IW 00. 0 ( C.) 0 0 foF->ON N DU 1 - W10 I U. '- 3 000000 NZ Z < N 0 3 u 0 - U0 0-r' 0 03 03 > Z 3 3 M -C> Ix O OO uI 0 Z4N LI. I--0D0 Q 3 UJ .. ( u-0 >Z'4 LUu 0 W > - C + + + 0("4 C',,- W 0 Z U. f 0 -~ 01 X 4 IL) (D IZIq) to '-11l O C0 .~X. uJ3,, 0+4+ 0 OWJ J"U) ('-C0D Or-(0m(N UU)IUW I- U 0 x * -- tM U7 - N4 0 3 uj 0"O or~.-cN - zo,,qED ( 0 z _l.J C) 00000000 0000 u 0 uL. N - 4 Wi.- .J J C-) a) uW 4 Z. N Z*uj Q Q' - 0 J .J Ui O WI-O W 0 W :E W I QI Wu. 0 U. (D 3 LlIW0 > I IIIIIIIIII11 I 11: I I >W a U U m 0 WO < L0 - 3 U 0 (-'U IL 3 > ~ u I I-W O. CU1 J J I (0N N - s00 0 wWI- N N - - 0000000000 " w Ir IO -U. I UJ 000 J 3 a.Q' u I _J - ',~ m c m m c m c c - -J > Z Z19 w F - I UWW - WYZ i -j i -j j -j i -j j -j Q CD Z Z U. z Z Z Z Z e - L 3 0 -'N N _ J-C 3 3 (J 0- 000 Ur 4 F- 0. I-- J333 J tL- =0. WI- I- O I- F- F- 1- Z wO. W I .J UJW U.3 J ILU.LIL ULW..J I Wi0. l W WY 0. IJ W 4> UL W W U. W W -J 0 W - O O Z W LI 010. 4 0OL0. Wi 0 a1-- 4J WO I ¢ 0 MU) L X Q IV -Y X Q)X W MD 0 CU - .JOO.JOOZ flW U) O - NI m X 0 J -j 3 0.:) J O 3X0.4- 3 - 3 c3N 3 F- 0. J 3 J 0 C u.XLU*J 3 _001 (N(0IM4 0 (NN 0 (0W -- 0 . 0 m 00 W 0 0 . Z Z I N(N ZIYz F- N1T O I3: m U U- U. II I Z QJ Cie-i UQ' I I 0 O N U- C) F- I W 0s CDCD 0 o N rl- - IV- 000 ---0-- 0 0 0- 0 .4 9t'' + 0. 00000o D coQ 000000i t- cl) t- u 0 O OLO o a b CD Or- cI 0) W 8r- 0 QOQ CDCDNOr-OI 0 oaoa w Ia=000 '4toM IV v -I- 07 0N 0 tiXC tD DC 0 0 N O v o Inr 0t0 t ~;N -ld0 ~; 0 i tD D T tOI- orI-tl-Goe'iLO 0 - (1 w0 r c t? ts N4 co0 r- N4-C ND 4 Go GoCN -4 N I t- mO OO O O OO c'- CY t O t 00000066 .000.0.0000000.'00 0. Ij C,)V) Xu

*LIL -L X 0 0C.I1 - 0 4 IV N tDN tN) to M - t 00 00000 0CO 0 cO 0 in x u 2 LO .+ 00000 I + + O x o +ll+1+ + 0 0 . C X z 0 0 M 0O(D 0 )0000 D d 0'- + cn 0 U) (ANN cc UJ 0 IU7(0 - V N- o U UiI ZI I-- it _1 0 0 v n VI) n ) 0C 7) - CD)'V 0 Q co 4 Z 0 I1t 4 co a {8 0)O T a: tr (1 0) D 0 o 0O S r- - . U. z LUJ . * CD () c T r- M r- r- M'U '- U7CDXL < 0.CZ 3 1 0 -J 0'* -N Cl N q 0 OO 0 +4+ ...... I 17 c7 1 000 -J * 0 1 (DOco LO t C 0N4 LUJ 0 0 00 O 00000 0 O 0' -N UI 0 0000 U .M, ... x0 0 Z IONU33) C 30 t 4 _x ww W - CYI ..JU 0 w C x 7 CZ co V W1- LU LULU 0 Lu 0. o I N>u. LUJ z II I r r U -J =-i QjQi -j -j -j -J z OW IJJ-J..JJJ10000000 1 I 4 .,.-~.i- co00aN c '0. N 000a oc N s0 0 s .-j _ DN N N Nzag Z Xn x Ul ai .m m coaoaoacac U)1-0(0 0(0 0(0 0(0 0U)- w -3 JNI 11111j- - 0 -0 0 Il wI-0NL 0 O*(1- LU - Cw z C z 4>0 3 3 31 3 W - U) LU)4 U) 00 3t 0 30 > J) t asoo II . J 33 330 0. 33 33 0J 0. i W400.3LULU 2 4N .-.. JO 300300.1 u. ~3 .j U. -j 0 0.J..J0aIOJU. Li. 3003000 J- 0 3r .J0 ff0DOONI- W 0 ULLU 4 -j 0 LUWZo - -J I-JI .j U.LU .. U. U. WAJ I U) O u>0CdI> U. 00 U 0 U. LJ LJ U. UJ J -J 0 W O U1-0O U.j -i LU.j -j 0 NMLU Z -UJ LUJ-j -j Lu.j -i OxaCZ -J 00U0Z -j U~ 4:J JU - Z~a 0 wwI 0 1 .JOO.JOO N LUJ -j 01- 0-zJJ@O 1- U) ("1xO 2 :W I U) 0O ONNONN r U .4 OI ej -j I : :E00 '00N0 U)zn tnU) 2 N N (D O V) Un a: O OC 2 C '2 I3 C W U) LU - ~-tz: NOI N OOICININ O3Ia:4 UJ0 uuuZIIu U 0. N I UUUZII UUU . N 0 U, aLuI 4 0. U-U. | 000000000 0in000 0 0 M 0 r- CD 0 000 00000 aO (Ot' 0a 'N 4 N N -- No N Lu Lu I w.. 00;0808w N0-CN (D 4 J 0) . . . ., ... il

o O o* I- 0 z * o V".Ve. 0 oLu o u0O 0OzCa_ 4I. 'A '-0 'tNU DNt 0+Oa NZ zU -0 .4 I *O (ID -J ow tc'O aY) Vb N U- O .J u CZ ~W- Lu. 0 0 080000 vm O Lu z 0 OO OO0m It cflI U} 0 Lu . t . .__ w I0u~ 00Dt-c + a30 Z 0z z ,-m0 CY) N WJOU. I-, saw in in z S) m 0 z4~ . N -Jul- Lu Lu (AI- I m Ix CZ 0uj * z N. z * zw -J 0 4 0. zUJ 0 Lu.J 0 -J 001'- z .-u 4 I - N.> 0 Ulu Q)I 3 0z ~Lu U 0 -j i- ' n - C NtI I- I U- z O~~m~mtD O 0 N O0N T c N(n NO I: m a- mff -j C Z IC zCZu -j z LuJ 4 0 0 U 0 CD

0 QLAJw

. 000

NsW00 '-4 LI)

LU LW) LI)4J

jI4 -J-

sn 40tLq W 0 ~~~ ~ ~ ~~~~~~~tsn( 0ZO .. Lu 3 0 0.VN) X Z S ~ ~ ~ Z .0. 0 0 4.-eLI):. W I->v z z naX:E D :Ez l I-0

> ~ ~~~~~tn Q~~~~uQ uxu nu xu 0~~~Vn tr Z IL,0~cc:E OXt u) )uZ n0v Mn W- 0.xW Z Z L Wn z O ttOIf°: o LI)t O 0. 0 0L tJ O.O C~C) O-O me- W

CNC14i 0 x000 -X > X X XX IxO DZ.XXX - zX__-X 2 E _IC- > > A > Q. 0 I I : O L MYO OU£

XxXXXxx XX xJX -j > j > -i > -i ) > _X> -J ^ 0OZZ ICon x on> > ExEYxCc I-- NW - U) tr CN N10 0 LL I _- - > J X --- XXXOOOOw ZE W W I 0 X ..x x I > X X 0 0. Q 0. = IU)U)>>xYY 00U) . 40~.0 C, x x WO x x X > J > J »xxxxxI>Z > i 'O0>> 4-U)-J D > J X X ) 0.0.0.1 XOa>w>o>ZZY (uAJ W- IQ 0 -i U

LLJ 0cN Z -XOO-o o - X 4 N W 0100 0 0 0000 0z zI x)xx 0000000>-00-- 'J ( X X -x N -0 o o o lxxx J I 0O 0ooo00>>x 0 Cq - 5I XU ) o ) tmN UOU N) U - U 0 X X X mJXO > :E xE > U) 0> z Y xO O> O z Q O( U) U)0. 0. U)0.I IC a ~x 80> Ul D.I I Xj W X X 4 O z XXX zE x I-, 0. -xX X XX X X ZEX XX > _J)xx)x x(x z U)LL - I I I z- zC > > > > I 0.W 0.11U>- J :> cnonu>xx Iaxx > J IU)U)U)}U))I) U)U )U) V) 4 0. C u LA. 4 v awt- r L 0. I.U. 000000 0 000 L. UI V 0 O OQOIi 0 0 C4, | co 'm ; 0 000 OCO O LU r- O rq eq- 00 w 0 4 0 00 t-4't-N )t 0c')0 w tD OD I-4 I- 00C,,0cu,- N 0 t ) 0 0 -? JM- 1 ; NO 0Oizn N 000000000 00- 0 ..J . . . .! . . . -J 4C O I I I 8 0 LUI 8 m, 0 Q'uC o . 0- c. 0 6 rI

CD 0 O4 0 0 + , + 0 O 0 00)!-0)112!- 0 o 0 0 m) t-F N 0 * I.: al o O0 I- N w.0 * r--O0 - Dt- CD -0 8 -- 00 0 *Oi LO z *0 r.t *1*ao -( .D0 .O '- t tDC, _-(90) ; c- r- -t- to 0 w O-m C C')(N *1- . _ ., V-( cN O t 000 (N IO. 0 F IV N I-- - - . Lu '- C, u Z l -~0 · NO . ,* 4z A z -J 0.0l~- *~~~~ L. U * * ceQ=U CD0 0 mA m * ·- 0 -'- * *1'- 0) .)mO '- * ..JU) LAU o- 0 0Z I + O 0 0 0 0 0 0. CZ CN co '4- 2: C- 0 0 Y w P- '- DQ J ICZ ,I4 -W I: ZZ Ir'-c O .O LO 0 4 O 0 OOO '-'-00 !- ( -tco 4 0 0' r- Z Z ' .w t'-0 C. 0 t- ,C D tn4 co U Lu O a o oo- r o r- W,00 . . C- w w cI lz P - D r . r- _ w wD C CZ U. IVID0)C- m . C- '- I- D 0 CoC o -' rn ( N O r -o C, c, (N 0 IV 0 -Jo z 2 _ co . - N4 '- '- 0. 0. Z : 0 I'm 0 0 t0q - - c - Lnw - t snzoZa Q ZeS* I z -a'- * - '- . 4 0 I 4-, 4et - > 4 I-0 u 0 -Cn iun I n o * m Z -Y Z zr - C.) U)n -j Z U. 44 LU 4C 4 * in0 C -j 0 -J X0 i- s * LU-4Cr 0 * - w -s 2< a X LU * Unfr-o 0 n 0 *a., U LU LU LUXn I--Z U in C-- z U w0 . * z 0. * Z LU4 U. U. O * 0 * w . * .. O * Z 0z CZ - UQ CZr -X X I- U= Z, : - Zi-j0 0. 0. __ " w- x cc ettn o wUOO 0-4 Z O0 Cr '- U.U.-z-x EnZ 0 (S ix z > O _-i 0I Qe Z-i 0 ZL "I2 zIxIi' ., -D LUCrY I OLtu J 4 z 0 u. I-m 0 - U..J C, U,i~ .. 0 - X -j - U)U)Cos I -E -J LUZ ..O IC r C mII- 0 0 4 0. z 0. z t " - LUIC 0 0 0 0.D (n aZ = 4 . co 4 U)02:C') .. coU)I U)I-Cr0. Q z 02,i IO !.I xOL0- -' w n I-z ~~ J 0. cLU4 J Dc0 L Z 4 - CS: IAI%' OO4 T 0 c lz OZ> UN NO N _. ~ ~ i-O 0 0 x "Z "ID2: -o tZ LU0OD Cr w9ZLU " V) n > > : 00022:2:0 ELUt J E -U ILUI-U)( I Z w 2 I-LU H> < =0. 0.C, z 0.LULU.4 0 w -j 0 O~i 4 0t cr 0 (n,-> x_- 0. z-4 IE uJ 4 L- 0. inCDDCDO'- W IV~ ! M (1O N r- (0(0CDV ! I -D ( MC t D Cv,'-C C(N o. . .. to 0+O (N- 0( LUJ t-0 r Wn C IV W - N 88o.00 t'-Oo I- -J 000000000...... * .DO LU. 0 N - - N - N : n,, t ._ 0000000 O/QOOOO ZDFaaO o0 (NCD 01 - N N 0 t r( t (N 1 0 N0) ' 0 -10)00)IT .... c 0 0D11 t- O O 0 r- C. .0)i * r- (D O o-00r-00 (D I- (D z * :3 -00.1 40 .. ---000 .0(D 0 - (D CT f ...- _0(0(0 O t'-'- t N LL -q V 1 IV-- r- . Z - 0 -T LU NO 000 0000 z I.-. 0 00 -J4 0t D Q J.U # 4 tD-FO U 0 C0 ODL r n 0 0+ U) Y 0 0 VD-M CO LLU LUJ IT . . U't-0 cv) M' 0000 (N40) LU U- 0 0j cl o ( D z zn J .CN 10t-CC' - (0)i DU) Z-Z- oo @ a0 u)Cm -J · . .0-r- (0D0( 0C 0 0 .- 0 *-(') CD_D 0 (DT tAn z(3 000 000-tD ')C _'-'-0 0 4 zP-. tt- ISXN It D_ _ '-CN( NCD 0 0 0 N O J o U 4 I t ) LUO 0 I * II 0 wA 2 -, 0 f000 -N U) -i'm- Ufl) -J * -. -CY m C4 M 0 o aw M 4 -dU-;-- 11 -N IC- e I I -. L, LU, 0 ,O ) IT: 0. 2 0 0 I.= 2 0000 W I-U) nx CZ 0 (fQ.~ N(-CN Ci i i * z X4NU.O C._- U X * zU .OX I 0 LU U4L X LU Q - I 0 L. 0.O O0 0~ (92 co a x x120 U. U0. 3 (M I- LUI 0(N00r(N0 I I.4 -N (.O -N 0. U. I Z 4 Q Z m I 0 :) 0 U) I -AJ LUJ U0Xmt) I- zi I 0. tnU) 1-" 0. iQ Z x 't >- 0 0 (N00 z LU .1 m D 2 0 I. 0 I '- LUICZ>" < Q.-r 4COD 0.I I- .dJ!n l-l. .. I- 0.1-C Ia U) LU.4U I (O0 .a lU LU I i z I ( -j u)~- OI0C. II- U CN N -J Q 40Q XLU D -J- Z Z 1 a (N(NC0 - IuoOIC'4Z(I U 14 U 2 z W 0. Q.C0. 0 m x U 0. (OCW) U -I I-- 0 C-, 0. LU - U) z 0= 'r~ Z IQ0. 4 LU ) - 0

0 I-~~~~~~~~~~~~~~~~~~~~~~~~~~~W0 0 o 0

0 0 z I 0 Z I.- zzz zz 8 zzz zz 4 z " - - b . - ao NOoLn LI Cntn o i /n (nv O -'- 101010 10Lflau- 101010 1010u I : Q 1 * :E'z- s" a :EO N0E ESzI--a - --C: 0 11 UwU * 3 inJ0 z Iq 0 -j wnCD I -

V 0 Q ~)Ij 0. ..JO Z Z Z 1 0 uj I 0 -0 ZN uC" O0 0 Lu4.AJ1- 2 3N * (NIt -- X -. 1 J 0. * 0I 0 1 L U u * L. L 0.

I _ 0 CZ - 4 0 O '2 U O LL0. - U. 0.* 0 .e) -- F J (A --FJ i _j 0 L CN U. Z 4 LL Z 4 0 CZX CN 0 0. 0 0.-a L L x . of -I.010 Lli -wCZ~ -20 w C) t) I. L w U" - 1- LL -1 z a 10 1. 0. < F ULLW0.- 4< 1LU m L. 0z MCZ _ o 0 1. 10 0. 10 > 0.0 0 0 ) 0.0n > LU) - F- XW0.>-2u4xZ1 121tL I z L0. >- z x Z CDn Z c0a12 m xw wax:< o uw wLa t I: Ulw a o V) D Lu SF0I-: > Y - m : Q < I- af > m if I- m 2 U V)e 0. 4 - 4 (n 0U. U z r uw 10 Lu C, 4 LL 0. IL L Go 0 a 0 N I CO0 0 Lu I-O >x 000000000 -J I

.- In O0 0 (N g8g~88-o +0 ·0 o.. o o.. .. t-o.. H . D co c - T C12MlU o * 0 . _-0- C4 - Q i 0 z N*-'- IV tD *, O 0-0C.- V O12D t- N NCv - 4Z t- - NO zZ U - 0l :T IwS - -CN04 .o0 -J 0 C-i 4 LA.LuI m J tU C II 1 U II II 11 H H H =dlY _. z zzzzzzzzz + C]C0 N 0 uj QOC z , LU J aQ4- CZI- Q I- 1 1 F- 1- I F- F- F 4 0. z 0 -.I-.1-J-j z LZ ZZZZZZZZZ U) I- Ir - - ix z zzzzzzzzz w o m0LU I 0 0 000000000 00 v o O I. .J 0 1 n10. Ila. z In 0 110.0000.0.0.00 LuC4 z EXszzzzzzz Xz z > x I-<4 40 4000000000 I-J l0 a.4 WuC Luuuuuuuu z u 0 0 CY I- I-- L0 I-- In In U) z N

LU CD U- L CD 0. I.i. 0 0 t3- ,,. U. 0 LU 0N. I-- - 000000000 ..0 z -J i~91 I Wu i I ) CZ 0 '0"- U) 0 ("("C.), 4- 0C) +Go (D r- 0)00) *1 I'-CN 0 0 1 U1 - -CN -co z *0 . Iq'M Cl' to"- 0 IT0) CV aW4 -* cO LUO 0)0 -' - 0 0 I-- -Z 0W -T 0 _C _ NO z4 0 * . r- L/. -J _q-*edU)'4~. . @ ~ aa ~ 4 qU, .C4 0 *9 -C ul >.. I {D~~~~~~~~~I 0) U WNL -t Y 0 0 l~~ ~ _ WO Z' 1 I CY I- 11 N 0 I- LUi M 0 - 4 U) N M IV C'2CU) (0"- + 4 W m Z Z- "-0 '-or-- N ,(N'-N'- --4 W n Z' -130 a O . -Un *(O [IJ _ 1 D O r- ."- a . 4 z - I' 0'-0 I.- 0oU o _. Vs Z4 0.J - 4(4)n IS) WO 4- I I -~< 4 '- NN U) 5 4 4 U)U)Z Z * 0 0 Z * LU LU-LJ U - nI* W ,-4. * _- D Z * )1-0 0 LIU U) * U) C Q Q' O U) 1 00 ·. 0. * Z * eq I z P.UZN "I 0.I LQ LU 00 * LU _ 00- U) 1- *- CNIL * Z CO 1- M Owx * a. 4 '-4 0.0 Z X tll I 0 -o .3 4t )(U)1-3 0. vI 1 44-_ 1 -LL I. C 0 N4 4 Z ( Z zLUJ LU LU O W1 I--4U. . z I Z V) 0 4- O LU Z U)I W- -r :- :E QU0F z W s Lu 4 0. U)I-WLOLU 0 CK J coQ z W-CX W W O 0 .4Q 4 1 -0 U) 0. ZZ I nli-- LUj F- U)>- CD 00 LUW( I 0.WLJ JIW -rU) 1.- Z CN4 )(n LU 0 < .1-0.. 0 wN'OOI(NN(N ID -OJ - _ LU LUCZW VZI--0CZ an, *uU s z 4 LUJQ .4 .- LU >XU0U21X LU 4-eULU.. z LU1 0>40Q 2 0U'->U LUJ CZ J 1-U0X 4 0 3OQLUXZ 0. UJW > -C O LUJ U 30. Q50. .U X 2 4, LU CD 0(D 0000_ - C4 0 ? I I I 0a IL (0 .O- 000a a a 0 I. 000000000 000 Ln M M (D M C t ',- 0 v .4 ~0 N ., (0 0 -N'2-N C. r- L -t- C 0)C1I 0 + - M O IV co1! I" . (N4(' 0) to Ii) M' C('2r ( IV L00 C ON ·c U1 I ( 0 _ ('(DN L ( 0 0 0 N - N~ - C] - 0 0 -i 1 0 1~0 *eJ( LU I I o C! 0).O.. '.0... CO I-- ~~~~~O )~~~~~~~~ OcO 0 I I 0 00000 I' ' LU I I ITI I-- 0 0 0) 0 00t 0 (N00000 0) In (O 0 + U) mCY am C'NCD (DLn CDco 0)0)021(202U1-N G) t2 CD t- co CY . I . . .. * r-N* MIV -£ 00a) U -4 amw- aw u N c t * . . - 0 CO - C o IV_, O'- . z * -02 0 .CU2 , *C'2(N 0 CD0 '-Q _ , '-. I- LU _00 IT LE)0 V N - 0 4 Z cM vr t - 0 U) NO 00 000 0 z 0 I.. _J * 0 c U 000 000tl- O I- 0N 4 * LU '-020wo 0 C '-0~ 0 * - U) - 1- ( o-- w - w 0 (0 * 4 C, O <' O 0C-(D . .-- . '.0 - j L~U LU X . . oco W tl IS N - fft cn w In N4 O0I- -LC t-"v N Cqs 000wm 4~ + LU J0 * A I Z Z . ,~ 0 Go~ ~L .C1 I- z I- Cjt3 U) u .I :-CO *'-0 W *IV - (D- I-- 3 .-O 02 N *4'in ' Cy N 04 I - U)Ul z iZ 0 M -0-O-t v t F 0 lw 0 0 LU -- N - C" N 0. Z -j I - 0 VqU - N N - t- - 0. -C0Om 0000000 0 z V .. I I I I I I I I I W .. o Io Ic 0 000000 U) : 0i 0 # 0 N 0 I( 0 ( U W (0 -4 U) -4 U) LU L 4 * Uo -j U- I Z* -4 U'~~~LU -C0 0 J M (N 0)'-'- U(' CN to (0 > X O0 LUJW 0 U.0. 0LUa Un * U) 0 U .. 1L U. IL ) I-- L ...... U0.O * Z 3 30 * LU m * Z -j4 O! CC.)- 0 * U.n u w zzI-@ I z0 LU 0 v Zn o I c iC-- O. (0 -4 = :) CZ 0 CI0 O N D I CNO Z U)01('2wU o Z -, tL CL I. O I o q U)0Cxu z i- u0-) > U. -. CZ OICNzC'd(0 (0LU 4-LDC) -0.0 3 CL iCL O- * a M}U 4 LU -4 a U. C N LU 0O+ 0000 LL. 0. '000 U- '-0 co W9 - (Nt- 0 0 0 I I C" ' N 0 0 co ~ (SJ MI0. - - 0) c0 C', Y * LUi 00 ...... t LU0 N- I- 0 -j 000000000...... 01 z 0 cI t- LU - - N - Vt t CZ d 0000000 Cw I I 1-.41 a m0 lv*O sNOa ZX- m vt- Iw mN 0I VNM-CD0WT0000000 W mOw _ N as) O 0olW- ,,...-, (N a1 .~IV r- N :a M)t I- r- C O cN ...... * I- 04CM vt 0 0 0 - co *0 * '-* CD0 o00 *r- - z I * 0) N O'-Cq * CD 110 S- - . CM 4Z 0 ' +IT NO 0O 000 0 z.) 1,. - I I I 4 1'- ('] W 0~00 -i * = U 00C,'0, 0 0 4 1--O 0CD I~)O 0 e1 O C O CD 0(00 IJn * IV a) ,00 In) toma LU) 0N D1 X CS . IDa 01- 00 -j 0 . J - , 0 0 4 ...... 0. *, I--I .dto z in LU Wr-NO- ·- (D mC) 0 4 I U0 I-) U, * 0 O r- co .O * z- -J CZ LUi-(00 0 -' !". 4 -j 0 0. X -W > r- *C" !' 0f li1 - U- 4 C 0 - 0) 0' 0 - l .(,DO] (0(0 I Z Zj-q o 0-C,, 0} ( D (D { . Z,--Z * C3 -JO LU mU 00 O0O 4 0 z . n LUOc a: I I l I eKz CZ U)i In (0 -j CK ,-1 00 0 a I) 4- -J ~01NO U1 00 D 00 -C LI) -L z ..= U * P.. 30 -4 4 * 0 l.L - N 0- - inMC N (DOE LU I- * UJ UWI .I n I- 11, LL 0. w IU t I M 4 U...... 0. 3a: tLcn 0,- ' (0( ( ( 01 z .. 0 * LU 1-0. ,-I In iCJ _. a: 0. 0 I-LU Z CN LU LUJ I a: (0j a n .a: LU.* I : LU0 z < I-- LI X- Z ,*W'ZW._1 0 izI_l 4 C'a: Zz LU 4-4 : zLL0LU C, ) 0 -J ' o I- u: LU -j a:LU Z 2 > O) (A < UJ X z 4 0 . O' I- .4LUJ0. 0 0iLU LU (a) .4 I-z Z Ln LU I I-In Lm U.J co a(0 I0. U. z z I- 0. I tz 0 4o( 0 ( 0 0 . L LUJ MLL 0. (D0 N IO C -j nI- I-U. I-- -J I I- UCN I a: .,. I-0 0(N0IN(N(NCDoI O Q (N(N(N (0I CZ LU CNI0 :) a: O NI oaO I N ( I- LUJ LU z>1 4 ULL -j a: z 0. 0.I--0 c c= N "ZII LLUJ 0. 0(a. (auauaZIIa 4 I- O--0. 0 ~~ ~~0 . X 0. U) L4 CZiz -4 00 I 0 in .4 0 x> X "0 C -4 -, Ln

LA. Ln --1g10N 0 t-000

I tD T I-- .4 000000000 -j ...... LU Ij I

n c- * r- 04 r- * 3 00- z * -at _ . . r- O In -(D 01 . Cl-M M oo N o uo0 · 0'-0o-- 0 * C.) t 0 LU '-' t 4 z c'~nS 0*,-0 NO z -j.. U.(j=U jW~ -0 C' tD r- t c 0 N 0 LUj Or~aor a: f-0 z z_ .O0 -ID co ob_ ..J _ CO ' t.- .C' IT I Z Z I '," (D- 0 -c . D D(D 0 (0 0 O. V)YU 0 N -(0( Cl UR z -J '-(00t'- IV 0 M !. .cw . _o -jo o (N .4 0 fLu0 2 .4'- .4 U) ..J LIU- 3 4 zn. -JC) L I- wiLUI I Lai U) WL~ n n I- 0..n I0 * z 0 * uJ 4 z _33'- 0 2a- I JI CN > -J 0 S N. Z x Z tL 0 LU I 0 *N -J C.01x I- 0 0. Z N z "OCIm m LU ~ i. I- 0 I- (0 z : co* en LU nI--' 0 LU C OOXNNNJ'.W mI J 1: U U Uz m m **> Z ~ 0- (Aes 0.> U < Z S lJJ 0z I-- (J 0 U) 0. M F . I.- ujWZ < C,

G0o 0 ,q. IN cn (NLO 00o) 0000 u -MI)0n LO w Ma) a)aw omr a M a co - oo8 16 . .6 .. O 43 000

C', 0 0C', 0z 0 ! Ca 0 C' 0 - O'z z 0 0 Z za Z 8 . - 0. 0. CL I (A